Biogeosciences, 8, 1415–1440, 2011 www.biogeosciences.net/8/1415/2011/ Biogeosciences doi:10.5194/bg-8-1415-2011 © Author(s) 2011. CC Attribution 3.0 License.

Soils of Amazonia with particular reference to the RAINFOR sites

C. A. Quesada1,2, J. Lloyd1,3, L. O. Anderson4, N. M. Fyllas1, M. Schwarz5,*, and C. I. Czimczik5,** 1School of Geography, University of Leeds, LS2 9JT, UK 2Institito Nacional de Pesquisas da Amazonia,ˆ Avenida Andre´ Araujo,´ 2936, Aleixo, CEP 69060-001, Manaus-AM, Brazil 3School of Earth and Environmental Sciences, James Cook University, Cairns, Qld 4878, Australia 4School of Geography and the Environment, University of Oxford, South Parks Road, Oxford, OX1 3QY, UK 5Max-Planck-Institut fuer Biogeochemie, Postfach 100164, 07701, Jena, Germany *now at: Fieldwork Assistance, Postfach 101022, 07710 Jena, Germany **now at: Department of Earth System Science, University of California, 2103 Croul Hall, Irvine, CA, 92697-3100, USA Received: 18 December 2008 – Published in Biogeosciences Discuss.: 8 April 2009 Revised: 9 March 2011 – Accepted: 6 May 2011 – Published: 1 June 2011

Abstract. The tropical forests of the Amazon Basin occur 2000; Irion, 1978). On the other hand, factors such as soil on a wide variety of different soil types reflecting a rich di- temperature and soil moisture regimes are common to many versity of geologic origins and geomorphic processes. We Amazonian soils (Sanchez, 1976; van Wambeke, 1978). In here review the existing literature about the main soil groups the early days of soil science in this region, Marbut and Man- of Amazonia, describing their genesis, geographical patterns ifold (1926) observed at least six different groups of soils oc- and principal chemical, physical and morphologic character- curring commonly in the region, suggesting that many soils istics. Original data is also presented, with profiles of ex- occurring in the tropics had little or no morphological differ- changeable cations, carbon and particle size fraction illus- ence to those observed in the temperate zone. A subsequent trated for the principal soil types; also emphasizing the high view did then, however, emerge that tropical soils are invari- diversity existing within the main soil groups when possible. ably ancient, lateritic and intensively weathered. Although Maps of geographic distribution of soils occurring under for- this view persists to some extent even to the current day, both est vegetation are also introduced, and to contextualize soils Sanchez (1976) and Richter and Babbar (1991) demonstrated into an evolutionary framework, a scheme of soil develop- that tropical soils are actually very diverse, encompassing all ment is presented having as its basis a chemical weathering different taxa, from the lowest to the highest pedogenic lev- index. We identify a continuum of soil evolution in Amazo- els. Indeed, by that time Sanchez and Buol (1975) had al- nia with soil properties varying predictably along this pedo- ready found that soils previously mapped as Ferralsols in the genetic gradient. Peruvian Amazon actually were Ultisols, Alfisols and Incept- sols (Acrisols, Luvisols/Lixisols and Cambisols in the World Reference Base soil classification system: IUSS Working Group WRB, 2006), suggesting that ancient Ferralsols may 1 Introduction actually be confined to areas of the Guyana and Brazilian shields. Sombroek (1966) also reported a large diversity Tropical soils can arise from a wide variety of parent ma- of soils in his studies of the Brazilian Amazon, describing terials, climatic conditions, biotic interactions, landforms, a high diversity of “latosols”, “kaolinitic latossolic sands”, geomorphic elements and soil age. Many of these factors “podzols”, “lithosols”, ground water laterites, hydromorphic vary more widely in the tropics than in the temperate zone grey podzolics, “Regosols”, Gleysols, saline and alkali soils, (Sanchez, 1976; Richter and Babbar, 1991). Amazonia com- Indian black earths and terras roxas estruturadas (equivalent prises a vast and heterogeneous region, with many of these to present day Nitisols) plus other minor and uncommon soil factors, especially parent materials; landforms, geology and groups not properly identified. geomorphologic history varying widely (Sombroek, 1966,

Correspondence to: C. A. Quesada ([email protected])

Published by Copernicus Publications on behalf of the European Geosciences Union. 1416 C. A. Quesada et al.: Soils of Amazonia with particular reference to the RAINFOR sites

Richter and Babbar (1991) gave an analysis of available soil surveys up to that time, comparing results from the FAO World Soil Map (1988) and the Brazilian Soil Survey for Amazonia (EMBRAPA, 1981), also giving estimated cov- erage areas for each different soil order. They estimated that Ferralsols covered 0.391 of Brazilian Amazonia, with Acrisols covering 0.323, Gleysols and Plinthosols 0.063 and 0.069 respectively, and with Arenosols covering 0.044, Lep- tosols 0.045, Podzols 0.028, and Cambisols 0.013 of the area with Fluvisols and Nitisols covering less than 0.010. They also indicated that there was a considerable bias towards the dominance of Ferralsols in the FAO map which was related to the methodology used by the FAO system; Ferralsols being mapped on the basis of estimates of climate and vegetation data instead of empirical soil analysis. Much of the soil diversity in Amazonia has originated from the considerable differences in geology and geomor- phology history that occur across the Basin. In particular, the Andean orogeny generated tectonic load and sediment Fig. 1a flux into lowland Amazonia, deeply transforming a previ- ously craton dominated land into the diverse edaphic mosaic of present day (Hoorn et al., 2010). Interestingly, much early work on this subject was not by soil scientists or geomor- phologists, but rather by limnologists interested in the causes of observed variations in the elemental composition of river waters within the Basin. Based on such observations, Fit- tkau (1971) divided the Amazon Basin into four regions, as shown in Fig. 1a. He considered the characteristically low levels of chemical elements (especially calcium) found in the waters of the core area of Central Amazonia (Sub-region IV in Fig. 1a) to reflect the already low soil fertility of this re- Figure 1b. gion (Aubert and Tavernier, 1972), this being associated with Fig. 1. (a) Division of the Amazon Basin fertility regions accord- the lack of geological activity in recent times. In addition, it ing to Fittkau (1971). (I) Guyana shield; (II) Brazilian shield; was argued that the sediments deposited in this area would (III) Western peripheral area and (IV) Central Amazonia. Study also have had a very low nutrient content, such as for the sites depicted in this map relate to soil profiles used as examples sands derived from the ancient Guyana shield, which is at in this paper. (b) Maximum geological age for Amazonia (from least about 1700 million years old (Fittkau et al., 1975). High Schobbenhaus and Bellizzia, 2001). topographic stability combined with continuous hot and wet weather has also already resulted in a deep weathering and leaching of parent material with lack of erosion over the gen- of Fittkau’s Central Amazonian sub-region IV (east of Man- erally flat topography eliminating bedrock as a source of nu- aus) are mainly derived from rocks and sediments from the trients for this part of Amazonia. middle Tertiary with a high probability of belonging to the Figure 1b shows variation in maximum geological age for end of the Cretaceous and thus likely to have experienced the different provinces in the Amazon Basin. For the area more or less continuous weathering for more than 20 million encompassing Fittkau’s region IV specifically, it is possi- years (Irion, 1978). Included in this sub-region, the so- ble to differentiate large differences in substrate age with called Barreiras formation was originated from Cretaceous- the region involving the Madeira, Jurua,´ Purus´ and Tapa- Tertiary sediments, derived from the erosion of the Guyana jos Basins being considerably younger than the remaining and Brazilian shields (Herrera et al., 1978). This formation of Fittkau’s region IV (east of Manaus) which represents has been exposed for at least 100 million years (Fig. 1b) and the much older Barreiras and Alter do Chao˜ Formations. has been subjected to several cycles of erosion, deposition There should be a difference in geological age of at least and sedimentation. Such differences in geological history 100 million years between these areas, with the Madeira, have resulted in the well documented contrasting levels of Jurua,´ Purus´ and Tapajos Basin formerly being under influ- soil fertility and pedogenetic development between these re- ence of the Pebas and Acre marine/fluvial systems between gions (RADAMBRASIL, 1978). 23 and 7 million years ago (Hoorn et al., 2010). The soils

Biogeosciences, 8, 1415–1440, 2011 www.biogeosciences.net/8/1415/2011/ C. A. Quesada et al.: Soils of Amazonia with particular reference to the RAINFOR sites 1417

Fittkau (1971) also found the natural waters of his periph- in the Andean region are often termed “white-water rivers”. eral sub-regions (Northern, Western and Peripheral Amazo- These rivers are in fact brown in colour due to suspended nia in Fig. 1a) to be significantly richer in chemical elements particles eroded from mountain slopes, but what is important than the central Amazon Basin with the waters of Western is that they have higher levels of all nutrients compared to Peripheral Amazonia being considerably enriched. Not sur- other Amazonian rivers, notably Ca, P and Mg (Furch and prisingly, these are also the areas in the Amazon Basin where Klinge, 1989) – although concentrations are still below the relatively fertile soils occur: This particularly being the case average when compared to rivers draining the temperate and close to the Andes where topography plays an important role boreal regions (Herrera et al., 1978). Nutrient rich floodplain in the maintenance of soil fertility through erosion of the soil soils result from sediment deposition from such white-water surface and exposition of the underlying parent material (Jor- rivers (the Varzea);´ and such soils have been considered the dan and Herrera, 1981). flooded counterpart of the fertile and well-drained terra firme Encompassing Fittkau’s regions I and II respectively, the soils of western Amazonia (Furch and Klinge, 1989). Exam- ancient, pre-Cambrian Guyana and Brazilian shields, with ples of white-water rivers include the Solimoes,˜ Jurua´ and their series of igneous and metamorphic rocks, are placed Madeira, all of whom have their headwaters draining the An- to the north and south of the lower Amazon River. These dean Cordillera with complex and varied lithologies (Gail- are the oldest surfaces exposed in South America, with their lardet et al., 1997). In the Solimoes˜ Basin the core of the maximum geological age ranging from 1500 to 3600 million Cordillera consists of a pre-Cambrian basement formed ei- years (Fig. 1b). Despite being very old, most soils have been ther by sediments, igneous or metamorphic rocks. Evapor- formed in situ by the weathering of underlying crystalline ites, dark shales, fractured carbonates and Mesozoic red beds rocks (i.e. not derived from reworked sediments), which ex- are the main rocks overlying the basement. Further down- plains the occasional occurrence of unexpectedly high fertil- stream, Tertiary fluvio-lacustrine sediments (Ic¸a´ formation) ity soils in these two regions. dominate the lowland portions of the Solimoes˜ River (Gail- Between these shields and the Amazon River (Fittkau’s lardet et al., 1997; Hoorn et al., 2010). In the Andean part region IV) occur strips of Palaeozoic sediments in which De- of the Madeira River, the main rocks are Palaeozoic sedi- vonian shales are represented to an appreciable extent (Irion, ments associated with shales and rare Cambrian evaporites. 1978; Hoorn et al., 2010). By contrast, western Amazonian The Jurua´ and the lowland portions of Madeira and Solimoes˜ soils (Fittkau’s region III) mostly consist of pre-Andine sed- Rivers drain Tertiary fluvio-lacustrine deposits (Gaillardet et iments from the Cretaceous-Tertiary period uplifted in the al., 1997). Pliocene. Despite some regional variation in maximum age By contrast, waters that drain the pre-Cambrian Guiana (Fig. 1b), most of this formation probably commenced be- and Brazilian shields are generally known as “black-water” tween 1 and 2 million years ago. For instance, at the neigh- rivers, with the dark colour of their waters due to high con- bourhood of Acre state, Brazil, a number of fresh water and centrations of dissolved humic and fulvic acids in various marine sediments occur as a result of the Andean orogeny stages of polymerization (Herrera et al., 1978; Gaillardet et and fluctuations of the sea level during warmer climates al., 1997). But some tributaries that originate in the eastern (Irion, 1978; Kronberg et al., 1989, 1998; Hoorn et al., 2010). and central areas of Amazonia, usually draining Ferralsols, Also, the western Amazon region includes large areas where are often “clear”, with transparent, crystalline waters (Her- shallow soils on hillslopes dominate. Thus active contribu- rera et al., 1978). Neither black nor clear water rivers carry tions of weathering of parent material to soil fertility can be appreciable suspended soils particle loads. They are also expected (Irion, 1978). characterised by very low nutrient concentrations – a con- Floodplain soils of western Amazonia are also much more sequence of the low nutrient content of the substrates from recent. Formed in the Pleistocene and Late Holocene, these which these rivers drain (Herrera et al., 1978). soils are not much older than 5000 years (Irion, 1978). Al- The Rio Negro is probably the best example of an Amazo- though Fittkau’s region III has usually been considered fer- nian black water river. The very low nutrient concentrations tile (Sanchez and Buol, 1975), soils within this region often of the soils that drain into it have arisen as a result of several contain extremely high exchangeable aluminium levels (Lips cycles of weathering, erosion, and sedimentation. All nutri- and Duivenvoorden, 1996); a consequence of the ongoing ents are found in very small amounts with [Ca] being notably weathering of high activity aluminium bearing minerals such low (Furch and Klinge, 1989). as hydroxyl-interlayered vermiculite (Marques et al., 2002; This paper provides an up to date review of the most Lima et al., 2006). important soil types of the Amazon Basin, adopting the As noted already, differences in geomorphology and soil World Reference Base (WRB) approach for soil classifica- fertility across Amazonia are reflected in the dissolved ele- tion (IUSS Working Group WRB, 2006). Distribution maps mental compositions of the waters in the rivers draining each of the main soil types occurring under forest vegetation and region (see also Herrera et al., 1978; Irion, 1984; Medina its respective cover area are also introduced using the World and Cuevas, 1989; Furch and Klinge, 1989; Gaillardet et al., Soil Information Database (ISRIC, 1995). We contextual- 1997). In that respect, Amazon tributaries which originate ize the soils into an evolutionary framework, doing so by www.biogeosciences.net/8/1415/2011/ Biogeosciences, 8, 1415–1440, 2011 1418 C. A. Quesada et al.: Soils of Amazonia with particular reference to the RAINFOR sites

Table 1. Soil area beneath forest vegetation in Amazonia. the WRB soil classification system, CEC values were ad- justed for organic matter content and expressed on clay basis. Further, although exchangeable Al had been routinely anal- Reference Soil Group Area (×1012 m2) Cover fraction ysed using the silver thiourea method (Quesada et al., 2010) Ferralsols 2.350 0.316 for selected Brazilian samples it was also determined by the Acrisols 2.154 0.289 more common 1M KCl extraction method (Van Reeuwijk, Plinthosols 0.648 0.087 2002). Thus, all pit samples had been analysed using the Ag- Gleysols 0.615 0.083 TU method, with a subset of these samples analysed by both Cambisols 0.418 0.056 methods. Leptosols 0.405 0.054 With the purpose of using exchangeable Al as an aid for Arenosols 0.200 0.027 Fluvisols 0.187 0.025 soil classification using the WRB, [Al] values obtained from Regosols 0.144 0.019 Ag-TU extracts were then adjusted to 1M KCl levels using Lixisols 0.142 0.019 an equation derived by the means of a non parametric re- Podzols 0.141 0.019 gression. This showed the two methods to correlate well Alisols 0.020 0.003 (Fig. 2a), but with Al extracted by 1M KCl tending to be Histosols 0.016 0.002 slightly less at low concentrations and somewhat greater at Nitisols 0.002 < 0.001 high concentrations. Total 7.444 1 Exchangeable bases extracted by NH4Ac pH 7 yielded comparable results to extractions made by the Ag-TU method (Fig. 2b) with the exception of some samples with very low concentrations which were more strongly extracted proposing a scheme of soil development based on a chem- by Ag-TU. In agreement with the similarities in extraction ical weathering index. As well as drawing on previous work levels of exchangeable bases and aluminium by the two of others, original data is also presented to help demonstrate methods (Fig. 2), soil ECEC extracted by NH Ac pH 7 and the diversity of Amazonian soils, both within and between 4 Ag-TU were very similar (Fig. 3a). Nevertheless, CEC val- the various WRB soil groups. ues determined by ammonium acetate pH 7 almost inevitably yielded substantially higher values than did ECEC calculated 2 Material and methods from the Ag-TU extractions (Fig. 3a). Again differences in extraction levels were higher at lower cation concentrations 2.1 Study sites and declined with increasing concentrations. Such differ- ences in extraction power relate to the artificial charges cre- The soils of a total of 71 primary forest plots are used in this ated on the surface of variable charge colloids at higher soil study, including forests in Brazil, Bolivia, Colombia, Peru, pH (Uehara and Gilman, 1981) with the differences in ex- Ecuador and Venezuela. A subset of 18 soil profiles is used to traction power of CEC and ECEC resulting in very different exemplify soil characteristics among and within each WRB apparent base saturation levels (Fig. 3b). Indeed, the calcu- group. The geographic distribution of these sites is shown lation of base saturation by NH4Ac pH 7 for the soils in this in Fig. 1a. Details for each site are also given in Table 1 study almost always resulted in values less than 0.5 and in the of Quesada et al. (2010) along with a map showing the ge- placement of soils in the dystric category despite soils some- ographic distribution of each site and their soil classification times having high levels of exchangeable bases and base sat- up to RSG level (Fig. 1 of Quesada et al., 2010). uration in Ag-TU. To investigate the weathering levels of soils within the 2.2 Soil sampling and laboratory methods dataset, the weathering index Total Reserve Bases (6RB) was calculated. This index is based on total cation concentration Detailed descriptions of soil sampling and laboratory meth- in the soil and is considered to give a chemical estimation ods are given in Quesada et al. (2010) but are also briefly of weatherable minerals (Delvaux et al., 1989). Soil samples summarised here. Exchangeable cations were determined by (0.0–0.3 m) were extracted for total elemental concentrations the silver thiourea method (Pleysier and Juo, 1980), soil car- (Ca, Mg, K and Na) by strong acid digestion using concen- bon was determined in an automated elemental analyser as trated sulphuric acid followed by H2O2, with 6RB equal to described by Pella (1990) and Nelson and Sommers (1996). [Ca]T + [Mg]T + [K]T + [Na]T, where [X]T represents the −1 Particle size analysis was undertaken using the Boyoucos total concentration of each element in mmolc kg soil. method (Gee and Bauder, 1986). For the purposes of soil classification, cation exchange ca- 2.3 Representation of soil profiles pacity (CEC) and exchangeable bases were also measured by ammonium acetate pH 7 (NH4Ac pH 7, Van Reeuwijk, To illustrate representative vertical profiles of exchangeable 2002) on all pit samples. According to the requirements of cation concentrations, [C] and soil particle size distribution,

Biogeosciences, 8, 1415–1440, 2011 www.biogeosciences.net/8/1415/2011/ C. A. Quesada et al.: Soils of Amazonia with particular reference to the RAINFOR sites 1419

1000 ECEC CEC 100.0 -1

100

10.0 -1 4 c 10 mmol kg soil

1.0 4 c CEC (NH -Ac, pH7) or ECEC (NH -Ac+1MKCl)mmol kg soil

1

0.1 1 10 100 1000 0.1 1.0 10.0 100.0 ECEC (Ag-TU) mmol kg-1 soil Extractable Aluminium (Potassium Chloride Method) c Exrtractable Aluminium (Silver Thiourea Method) 1.0 -1 mmolc kg soil

0.8 1000.0 "Eutric"

4 0.6

100.0 0.4 "Dystric" -1 10.0 c 0.2 mmol kg soil Base saturation (NH -Ac method)

1.0 0.0 0.0 0.2 0.4 0.6 0.8 1.0

Base sauration (Ag-TUmethod) 0.1 Exchangable bases (Ammonium acetate method, pH7) 0.1 1.0 10.0 100.0 1000.0 Fig. 3. Cation extraction methods compared. (a) Cation exchange

-1 capacity (CEC) as determined by NH4Ac (pH 7) and effective CEC Exchangable Bases (Silver thiourea method) mmolc kg soil (ECEC)Fig. 3 as estimated by summing extractable bases (NH4Ac) and extractable aluminium (KCl) together, both plotted as a function Fig. 2. Cation extraction methods compared: (a) relationship be- AgTU extracted ECEC (sum of bases plus aluminium) (b) Base tweenFig. KCl 2 versus AgTU extractable aluminium. (b) Exchangeable saturation calculated using NH Ac (pH 7) as a function of base sat- bases extracted by NH Ac at pH 7 as a function of AgTU exchange- 4 4 uration estimated with AgTU. able bases.

number of horizons in the fitting program adjusted accord- we used the equal area quadratic smoothing spline approach of Bishop et al. (1999) using a value for the smoothing pa- ingly. Note that in all graphs presented here, cation values rameter, λ, of 0.01 with all profile depths, z, standardised presented are for the Ag-TU extractions only. prior to the fitting of the spline according to z∗ = z/zmax where z∗ is the standardised value and zmax is the maxi- 2.4 Preparation of soil distribution maps mum depth sampled. As detailed in Quesada et al. (2010) soils were typically sampled over a series of depths viz The Soil and Terrain database for Latin America and 0.00–0.05 m, 0.05–0.10 m, 0.10–0.20 m, 0.20–0.30 m, 0.30– the Caribbean (SOTERLAC), version 2.0, at a scale of 0.50 m, 0.50–1.00 m, 1.00–1.50 m and 1.50–2.00 m. Where 1:5 000 000 (Dijkshoorn et al., 2005) was used as the ba- physical constraints prevented sampling to 2 m depth, zmax sis for defining soils classes and their spatial distribution. was taken as the depth at which sampling stopped and the Non-forested areas were excluded from the analysis using a www.biogeosciences.net/8/1415/2011/ Biogeosciences, 8, 1415–1440, 2011 1420 C. A. Quesada et al.: Soils of Amazonia with particular reference to the RAINFOR sites

principal RSG are also shown in the Supplementary Informa- tion. These show that the geographic distributions of forest soils vary widely across the Amazon Basin and are usually associated with large scale geomorphologic features. For ex- ample, Acrisols and Ferralsols occur mainly on the Brazilian and Guyana shields and in the sedimentary zone along the central and eastern portions of the Amazon River. Although Ferralsols are absent in western and south-western areas of the Basin, Acrisols do occur in this region, but are appar- ently limited to Tertiary fluvio-lacustrine deposits and other sedimentary formations near the Andes. Other soils occur al- most exclusively in the vicinity of the Brazilian and Guyana shields such as Arenosols, Lixisols, Nitisols, Histosols and Podzols, with the latter occurring in areas of the Rio Negro basin as well. Plinthosols also occur in small patches along the Brazilian and Guyana shields but are most common in

Figure 4. sedimentary areas near the Jurua,´ Purus and Madeira rivers Fig. 4. Basin wide distributions of soils under forest vegeta- (i.e. Ic¸a´ formation). tion. Map based on the SOTERLAC–ISRIC soil database (ver- Regosols and Leptosols occur along both Shields but sion 2.0, 1:5 million scale) and the vegetation database of Saatchi are most common in the proximity to the Andean foothills et al. (2008) for South America. mostly outside of the Amazonian border. Floodplains and areas along the major rivers account for most of the Fluvi- sols and Gleysols, such as the catchments of the upper Ama- vegetation map derived from optical and microwave remote zon tributaries and along the Amazon itself. Large Gleysol sensing data over the Amazon basin, at 1 km spatial resolu- patches also occur in the Araguaia catchments in Brazil and tion, capable of discriminating 16 land cover types and with in north Colombia, also spreading along the Andean bor- an overall accuracy of above 0.85 (Saatchi et al., 2008). The der in sedimentary zones in the Peruvian Amazon. Some following land cover classes were aggregated and considered other soils are mapped as occurring almost exclusively in as “forest”: closed terra firme forest, open/degraded terra the Andean foothills and adjacent sedimentary zone such firme forest, bamboo/mixed semi-deciduous forest, liana as Cambisols and Alisols. Indeed, the proportions of soils dominated/open forest, transitional/deciduous forest, sub- occupying the eutric lower classification levels (Quesada et montane forest, montane forest, closed woodland, closed al., 2010) are much greater in this zone. The coverage swamp forest, open swamp forest and mixed vegetation of each soil group is summarized in Table 1. Ferralsols swamp. and Acrisols alone account for 0.61 of Amazonian forest The sub-division of the soil classes provided in the dataset soils, with Plinthosols, Gleysols, Cambisols and Leptosols from SOTERLAC were aggregated into the following main accounting for most of the remaining portion (0.09, 0.08, categories to encompass all the variability per WRB Soil 0.06 and 0.05 respectively). All the remaining soils alto- Reference Group (RSG): Acrisols, Alisols, Arenosols, Cam- gether cover less than 0.12 of the area. We note that the bisols, Fluvisols, Ferralsols, Gleysols, Histosols, Leptosols, estimation of soil coverage shown is based only on soils un- Lixisols, Nitisols, Plinthosols, Podzols and Regosols. Other derneath forest, and that the area mapped is defined for a new soils were not mapped due to their limited coverage in definition of the Amazon border (Soares-Filho et al., 2006). Amazonia; these being Andosols, Solonchak, Solonetz, Accounting for the diversity of Amazonian soils, Table 1 Phaeozems and Luvisols. A Geographic Information Sys- in Quesada et al. (2010) lists geographic coordinates, coun- tem (GIS) was used to compile and carry out map algebra try of location and the identified soil type for the 71 sites with the reclassified soil and vegetation maps combined us- sampled as part of that study. This shows that of the 32 ing Boolean operators in order to generate a final result of RSG in the WRB classification scheme, 14 Reference Soil soil types on forested vegetation formations. Groups were identified, viz. one Leptosol, five Gleysols, one Fluvisol, 13 Cambisols, one Andosol, one Nitisol, 10 3 Results and discussion Plinthosols, two Umbrisols, 10 Alisols, one Lixisol, seven Acrisols, 13 Ferralsols, two Arenosols and four Podzols. A 3.1 Describing Amazonian forest soils: distribution and large variation within RSG was also found, with three lower diversity level differentiations for Alisols, two for Acrisols, seven for Cambisols, five for Ferralsols, two for Plinthosols, two for The distribution and coverage across each Reference Soil Podzols and two for Gleysols. Geographical locations of the Group (RSG) are shown in Fig. 4. Individual maps for the soils (Fig. 1 of Quesada et al., 2010) suggest our sampling

Biogeosciences, 8, 1415–1440, 2011 www.biogeosciences.net/8/1415/2011/ C. A. Quesada et al.: Soils of Amazonia with particular reference to the RAINFOR sites 1421

-1 -1 Exchangeable cations (mmolc kg soil) Carbon concentration (mg g ) 0 5 10 15 0 20 40 60 0.00 0.05 0.10 0.15 Al 0.20 Na K 0.25 Mg 0.30 Ca

Soil depth (m) Rock 0.35 0.40 0.45 0.50

Rock

Fig. 6. Profiles of exchangeable cations and soil carbon for a Hy- Fig. 5. Forest vegetation above a Leptosol in Venezuela (ELD-34). Fig. 6 perskeletic Leptosol (Orthodystric) in Venezuela (ELD-34). Fig. 5 to have given a reasonable representation of the soils of the Amazon basin in six different countries. suffix “Orthodystric”, exchangeable cations concentrations With a view to demonstrating the diversity of soils in Ama- were moderately low, reflecting both parent material fertility zonia, each of the 14 different major Reference Soil Groups and the small degree of development of this soil. Nutrients found in our study are now considered. Representative soil were found only in the top soil, where organic materials ac- profile data is given for each soil group, with additional pro- cumulate. The fine earth fraction was very coarse (< 0.20 as files given for soils showing significant variation within their fine earth fraction), mostly due to mineral fragments which RSG. As in Driessen et al. (2001), soils are organized ac- had not been broken down to a fine degree, thus granting this cording to the major factors conditioning their morphologi- soil with the qualifier “Hyperskeletic”. The soil depth was cal, chemical and physical properties. very variable, but usually not exceeding 0.25 m. In terms of vertical structure, the soil had a well developed litter layer 3.2 Soils conditioned by limited age above a thin soil horizon over rock. The diabase rock from which this soil originates is resistant to weathering and bare 3.2.1 Leptosols rock outcrops accounted for most of the ground surface, with the soil mostly appearing in concave portions of the terrain or Leptosols (some Entisols, Orthents and other lithic sub- in cracks among the rocks. The relatively high carbon con- groups in US Soil Taxonomy) are shallow soils over continu- tent of this soil seems to reflect the accumulation of litter in ous rock or soils that are extremely gravelly and/or stony. Al- the surface, as there is little opportunity for carbon to move though being azonal, Leptosols tend to be found on rock out- down in the soil, being thus ultimately conditioned by the crops and mountainous regions (Driessen et al., 2001). Here soil depth. they usually occur on rocks that are resistant to weathering, With the purpose of facilitating an understanding of WRB where erosion has kept pace with soil formation, or where the qualifiers for readers not familiar with this soil classification, soil surface has been removed (Buol et al., 2003). They are Table 2 lists a glossary with the meaning for the most com- characterized by various kinds of continuous rocks or uncon- mon suffix and prefix used through the text. Complete de- solidated material with less than 0.2 of fine earth by volume. scriptions of each qualifier can be found at IUSS working Within the Amazon Basin they often support short forest or group WRB (2006). savannas as in the Inselberg complexes in the frontier zone of Guyana, Brazil, Venezuela and Colombia, as well as in 3.2.2 Cambisols the Andean fringe (Reatto et al., 1998; Cotler and Maass, 1999; Sombroek, 2000). The Cambisol group (Inceptsols in US Soil Taxonomy) con- Nevertheless, Leptosols do occur under forest and one ex- sists of soils with incipient formation. The initial stages of ample can be seen in Fig. 5, here illustrating one of the high- transformation in the soil material are evident, with devel- est forest biomass in our dataset with many large trees over oping soil structure below the surface horizon and discol- almost bare rock. Figure 6 show the profiles of Ca, Mg, orations (Driessen et al., 2001). Cambisols tend to evolve K, Na, Al and soil carbon for this Hyperskeletic Leptosol from a variety of medium and fine textured parent materi- (Orthodystric) in Venezuela (ELD-34). Consistent with the als. They are also characterized by profiles of only slight www.biogeosciences.net/8/1415/2011/ Biogeosciences, 8, 1415–1440, 2011 1422 C. A. Quesada et al.: Soils of Amazonia with particular reference to the RAINFOR sites

Table 2. Glossary of WRB qualifiers used through the text. For complete description of soil classification terms see IUSS Working Group WRB (2006), report 103.

Qualifier Description Acric Soils having argic horizon, low activity clay and low base saturation. Occurs as prefix qualifier in soils where other diagnostic properties have prevalence (i.e. Acric Ferralsol) Albic Light-coloured subsurface horizon in which free iron oxides and clay have being removed or segregated, revealing the true colour of remaining sand and silt particles Alumic Having Al saturation ≥ 0.5 in some layer between 0.5–1 m from surface Andic Properties resulting from the moderate weathering of pyroclastic material Arenic Having a texture of loamy fine sand or coarser in a layer ≥ 0.3 m within 1 m from surface Argic Subsurface horizon with distinct higher clay content than the overlying horizon Cambic Subsurface horizon showing evidence of alteration in relation to least weathered underlying horizons. Clayic Having a texture of clay in a layer ≥ 0.3 m within 1 m from surface Cutanic Having clay coatings in some part of an argic horizon Endogleyic Having reducing conditions between 0.5 and 1 m from surface AND gley colour patterns in 0.25 or more of the soil volume Endostagnic Having stagnic colour patterns and reducing conditions Ferric Having an horizon with segregation of Fe or Fe and Mn Ferralic Subsurface horizon resulting from long and intense weathering in which the clay fraction is dominated by low-activity clays and the silt and sand fractions by highly resistant minerals, such as oxides of Fe, Al, Mn Geric Soils with very low ECEC Gibbsic Having 0.25 or more gibbsite in the fine earth fraction Haplic Indicates soils not associated to any specific or intergrade qualifier. Used only when no other prefix applies Hyperalic In Alisols only: soils having silt/clay ratio < 0.06 and Al saturation ≥ 0.5 Hyperskeletic Having less than 0.2 fine earth fraction by volume Hyperdystric Base saturation ≤ 0.5 throughout between 0.2 and 1 m from surface AND < 0.2 base saturation in some layer within 1 m from surface Hypereutric Base saturation ≥ 0.5 throughout between 0.2 and 1 m from surface AND ≥ 0.8 base saturation in some layer within 1 m from surface Melanic Thick, black surface horizon, which is associated with short-range-order minerals (commonly allophone) or with organo- aluminium complexes, typically originated from pyroclastic material. Mollic Well structured, dark coloured surface horizon, with high base saturation and moderate to high organic matter content Nitic Distinct clay-rich subsurface horizon with well developed polyhedric or nutty structure with many shiny ped faces Orthodystric Base saturation ≤ 0.5 throughout between 0.2 and 1 m from surface Orthoeutric Base saturation ≥ 0.5 throughout between 0.2 and 1 m from surface Ortsteinc Having a cemented spodic horizon

Oxyaquic Saturated with O2 rich water for more than 20 days AND no gleyic or stagnic colour pattern Plinthic Subsurface horizon that consists of an Fe-rich humus-poor mixture of kaolinitic clay with quartz and other constituents, and which changes irreversibly to a layer with hard nodules, a hardpan or irregular aggregates on exposure to repeated wetting and drying with free access of oxygen Rhodic Very red soils

Silandic In Andosols only: having one or more layer with andic properties AND Siox ≥ 0.06 OR an Alpy to Alox ratio of less than 0.5 Siltic Having texture of silt, silt loam, silt clay loam or silt clay in a layer ≥ 0.3 m within 0.5 and 1 m from surface Spodic Subsurface horizon that contains illuvial amorphous substances composed by organic matter and Al or Fe Umbric Thick, dark-coloured, base-depleted surface horizon rich in organic matter −1 Vetic Having ECEC of less than 60 mmolc kg clay in some layer within 1 m from surface Vitric Properties related to soil layers rich in volcanic glass and other primary minerals derived from volcanic ejecta, which contain a limited amount of short-range-order minerals Xanthic Very yellow soils

Biogeosciences, 8, 1415–1440, 2011 www.biogeosciences.net/8/1415/2011/ C. A. Quesada et al.: Soils of Amazonia with particular reference to the RAINFOR sites 1423

-1 -1 or moderate weathering, and by the absence of appreciable Exchangeable cations (mmolc kg soil) Carbon concentration (mg g ) Particle size distribution 0 20 40 60 0 5 10 15 20 0.00 0.25 0.50 0.75 1.00 quantities of clay illuviation, organic matter accumulation 0.0 or readily extractable aluminium and iron compounds (IUSS 0.2 Working Group WRB, 2006). Most Cambisols are soils with 0.4 Al 0.6 Na K some horizon differentiation and are in a transitional stage 0.8 Mg Ca of development; actively developing from a young soil to a 1.0 Clay 1.2 Silt more mature form which will have an argic, spodic or fer- Soil depth (m) Sand 1.4 ralic B horizon (Buol et al., 2003). The Cambisol group 1.6 keys out late in the WRB taxonomy hierarchy which implies 1.8 (a) that this group may include some soils that just missed out 2.0 one or more requirements for other reference groups (IUSS Saprolite Working Group WRB, 2006). High quantities of weatherable

-1 -1 minerals and the absence of any signs of advanced pedogen- Exchangeable cations (mmolc kg soil) Carbon concentration (mg g ) Particle size distribution 0 25 50 75 100 0 5 10 15 20 0.00 0.25 0.50 0.75 1.00 esis is further evidence of the early stages of the soil forma- 0.0 tion characteristic of cambisols (Sanchez, 1976; Buol et al., 0.2 Al Clay Na Silt 0.4 Sand 2003). There is however, often evidence of initial weather- K 0.6 Mg Ca ing and transformation of primary minerals (Sanchez, 1976; 0.8 Buol et al., 2003). The cambisols of tropical regions gen- 1.0 1.2 Soil depth (m) erally occur in recent geomorphic surfaces that may be ero- 1.4 Rock sional or aggradational (Driessen et al., 2001). Given the 1.6 1.8 (b) high weathering intensity of the tropics, young alluvial de- 2.0 posits and fresh rocks exposed at shallow depths may evolve to cambisols in a relatively short time (Buol et al., 2003). Rock Most cambisols in this study were found close to the An- Fig. 7 des on erosional surfaces in western Amazonia, but this soil Fig. 7. Profiles of exchangeable cations, soil carbon and texture type has also been reported to occur along the Solimoes˜ river for (a) a Haplic Cambisol (Alumic, Hyperdystric, Clayic) in south floodplain (Moreira et al., 2009). Peru (TAM-05) and (b) a Haplic Cambisol (Orthoeutric) in Ecuador (BOG-02). Figure 7 shows two Cambisol profiles of contrasting fer- tility: A Haplic Cambisol (Alumic, Hyperdystric, Clayic) located at Tambopata, south Peru (TAM-05, Fig. 7a) and a loam, this being considered typical for this type of soil. Soil Haplic Cambisol (Orthoeutric) located in Ecuador (BOG-02, depth was shallow, reaching saprolite at 0.8 m and rock at Fig. 7b). For the Peruvian soil, exchangeable bases were 1.1 m. This soil was strongly conditioned by its steep topog- very low, declining strongly with depth, but with exchange- raphy being constantly renovated by erosion. able aluminium concentrations high, and increasing slightly As also noted by Sombroek (1984), the above compari- with depth. Soil carbon was higher in the soil surface due to son illustrates the importance of appreciating that Cambisols organic matter inputs from the forest litterfall, whilst as for are highly variable in terms of their fertility. This varia- most profiles examined as part of this study subsurface car- tion occurs because their close proximity to parent material bon declines slowly with depth. Clay content was high, also makes them strongly dependent on quality of the weathering increasing slightly with depth, but in this case such an in- substrate. Also, varying degrees of weathering and mineral crease in the control section does not fulfill the requirements transformation are observed, adding to the variability found for argic horizon such as is the case for Alisols, Acrisols and (Buol et al., 2003). Another source of variability in soil fer- Lixisols (Sect. 3.3.3 to 3.3.5). The elevated aluminium pro- tility may rise from the inclusion in this classification group portion of this soil and its high clay content confers the suffix of soils that just failed criteria for other groups (Wilding et descriptors “Alumic” and “Clayic”. This soil had a maximum al., 1983; IUSS Working Group WRB, 2006). Although this depth of 1.7 m below which saprolite was present. The verti- did not seem to be a significant factor for this study. cal distributions of clay and [Al] suggest that transformations towards another soil type are currently taking place, with this 3.3 Soils conditioned by a wet tropical climate soil most likely evolving to an Alisol. Suggestive routes for soil development are discussed further in Sect. 3.7. 3.3.1 Plinthosols By contrast the exchangeable bases for BOG-02 were among the highest found in Amazonia (Fig. 7b) with very Plinthosols (Plinthic Great Groups in US Soil Taxonomy) low extractable aluminium content, and cation concentra- are soils which have as their principal characteristic the el- tions varying little with depth, this granting this soil the de- evated concentration of plinthite, an iron rich, humus poor nomination “Orthoeutric”. Carbon profiles showed a sharp mixture of kaolinite clay and quartz which changes irre- decline in depth. Soil texture was sandy clay loam to clay versibly to hardpans on exposure to repeated wet and drying www.biogeosciences.net/8/1415/2011/ Biogeosciences, 8, 1415–1440, 2011 1424 C. A. Quesada et al.: Soils of Amazonia with particular reference to the RAINFOR sites

-1 -1 cycles (Sombroek, 1984; IUSS Working Group WRB, 2006). Exchangeable cations (mmolc kg soil) Carbon concentration (mg g ) Particle size distribution 0 20 40 60 80 0 10 20 30 0.00 0.25 0.50 0.75 1.00 Plinthite most commonly evolves from the weathering mate- 0.0 rial of basic rocks as opposed to acidic ones. As formation 0.2 of plinthite is associated with fluctuations in ground water 0.4 0.6 Clay levels, these soils are thus often associated with lower land- 0.8 Al Silt Na Sand 1.0 K scape positions (Sombroek, 1966; Lima et al., 2006). The Mg 1.2 Ca Soil depth (m) development of the plinthic layer depends on the accumula- 1.4 tion of sesquioxides through removal of silica and bases un- 1.6 1.8 der hydrolysis, following the discharge of weathering prod- (a) 2.0 ucts. This results in a relative accumulation of weathering Compact Clay resistant materials such as sesquioxides, quartz and kaolin- Layer ite (Driessen et al., 2001). Absolute accumulation of these 1 -1 materials is also possible through alluvial or colluvial depo- Exchangeable cations (mmolc kg- soil) Carbon concentration (mg g ) Particle size distribution 0 20 40 60 80 0 10 20 30 0.00 0.25 0.50 0.75 1.00 sition. Another mechanism for plinthite formation is the seg- 0.0 0.2 regation of iron which occurs under alternating reduction and Al 0.4 Clay Na Silt oxidation conditions (Sombroek, 1984). Under water satura- 0.6 K Mg Sand tion much of the iron is in the ferrous form and therefore 0.8 Ca 1.0

mobile. But this iron precipitates as ferric oxide when condi- Soil depth1.2 (m) tions become drier and does not re-dissolve, or only partially 1.4 1.6 re-dissolves, when conditions become wetter again. In its 1.8 (b)Continuation of C Horizon unaltered form, plinthite is firm but can be cut with a spade. 2.0

If the land is later uplifted or suffers changes in its moisture Continuation of C Horizon regime, however, plinthite can become irreversibly hardened Fig. 8 to form petroplinthite. Fig. 8. Profiles of exchangeable cations, soil carbon and tex- Soils with plinthite layers are common under tropical for- ture: (a) for a Haplic Plinthosol (Alumic, Orthodystric, Siltic), at est vegetation, and soils with petroplinthite are most com- Acre state, Brazil (DOI–02) and (b) for an Endostagnic Plinthosol mon in transitional zones between rain forests and savannas (Alumic, Hyperdystric) in Colombia (AGP-02). But petroplinthite can also occur in wetter areas where most of it has presumably been dislocated, hardened, transported and finally deposited as alluvial or colluvial parent mate- (Lima et al., 2006). Carbon does not present any special pat- rial. Despite Plinthosols being found, at least to some ex- tern apart from being relatively high in the surface layer. tent across all of the Amazon Basin (Sombroek, 1966, 1984; The second profile is for an Endostagnic Plinthosol (Alu- Sombroek and Camargo, 1983) most of the Plinthosols iden- mic, Hyperdystric) located in south-eastern Colombia, west- tified in this study were in western Amazonia, where they ern Amazonia (AGP-02, Fig. 8b). This soil had a slightly commonly occur associated with well defined geomorpho- stronger influence of slow draining water than for DOI-02, logical and/or landscape characteristics (Fritsch et al., 2006). with the presence of stagnic colour patterns in the profile, Physical characteristics of Plinthosols are usually restrictive; similar to that described by Fritsch et al. (2006). Even the soil is very compact with high bulk density and the soil though exchangeable base concentrations were much lower structure generally weak for all horizons (Sombroek, 1984). than DOI-02, the overall distribution pattern was virtually Two soil profiles have been chosen to illustrate lower level identical, but with higher aluminium concentrations through- variations within this group. The first soil profile is a Hap- out. Carbon content showed a similar pattern as for DOI-02 lic Plinthosol (Alumic, Orthodystric, Siltic) located in Acre this also being the case for depth dependent variations in soil state, Brazil (DOI-02, Fig. 8a). For this soil, exchangeable particle distribution. Ca and Mg are relatively high at the surface in relation to other Amazonian soils, but with Ca declining sharply to a 3.3.2 Ferralsols depth of 0.4 m, below which it remains relatively constant −1 at about 10 mmolc kg until 1.2 m, then declining to prac- Ferralsols (Oxisols in US Soil Taxonomy) are soils that carry tically zero values. Magnesium follows a different pattern, marks of strong weathering and desilication and typically oc- being mostly constant until 0.4 m deep and then sharply in- cur in tropical, humid, free draining environments. They creasing until 1.2 m, after which it starts to decline to prac- generally cover old geomorphic surfaces or develop over tically zero values at 1.7 m. Aluminium on the other hand, sediments that were pre-weathered from ancient regoliths is practically absent until a depth of 0.3 m, sharply increases (Buol, 2002; Buol et al., 2003). Ferralsols may also de- until 0.7 m after which it increases again below 1.2 m. This velop in younger materials which weather rapidly, such as vertical distribution of cations in Plinthosols profiles mostly basic and ultrabasic rocks and geologically old volcanic ma- reflects the ongoing removal process of silica and bases terial, but these occurrences are generally of limited spatial

Biogeosciences, 8, 1415–1440, 2011 www.biogeosciences.net/8/1415/2011/ C. A. Quesada et al.: Soils of Amazonia with particular reference to the RAINFOR sites 1425

-1 -1 importance (Buurman and Soepraptohardjo, 1980; Beinroth, Exchangeable cations (mmolc kg soil) Carbon concentration (mg g ) Particle size distribution 0 10 20 30 0 10 20 0.00 0.25 0.50 0.75 1.00 1982). Ferralsols are commonly found on stable topography 0.0 of tropical regions where wet and hot climate favour inten- 0.2 0.4 Clay sive weathering (Sanchez, 1976). As reported by Driessen et Silt 0.6 Al Sand 0.8 Na al. (2001), and because of their extremely advanced state of K pedogenic development, the weathering of the mineral frac- 1.0 Mg 1.2 Ca Soil depth (m) tion of Ferralsols releases only negligible amounts of nutri- 1.4 ents or aluminium. Moreover, P adsorption capacity is high 1.6 1.8 (a) (de Mesquita Filho and Torrent, 1993) and the total P pool 2.0 is usually small (Smeck, 1985). Also, no neo-formation of

-1 -1 clay minerals is expected to occur. This mineral fraction is Exchangeable cations (mmolc kg soil) Carbon concentration (mg g ) Particle size distribution 0 10 20 30 0 10 20 0.00 0.25 0.50 0.75 1.00 usually dominated by kaolinite and iron/aluminium oxides 0.0 as goethite, hematite and gibbsite (Sombroek, 1984; Buol, 0.2 2002), a mineral assemblage that makes Ferralsols variable 0.4 Clay 0.6 Silt Al Sand charge soils (Herbillon, 1980; Qafoku et al., 2004). Despite 0.8 Na K 1.0 Ferralsols being chemically poor and with an acidic reaction, Mg 1.2 Ca Soil depth (m) most have well developed physical conditions. In general 1.4 they are deep, well drained soils which have low silt content, 1.6 1.8 low bulk density, strong fine and very fine granular structure, (b) 2.0 and a high hydraulic conductivity (Sanchez, 1976; Richter and Babbar, 1991; Buol and Eswaran, 2000; Buol, 2002). Fig.Fig. 9 9. Profiles of exchangeable cations, soil carbon and tex- Clays do not disperse in water and there is little osmotic ture: (a) for a Geric Acric Ferralsol (Alumic, Hyperdystric, Arenic) swelling because the cation concentration at the surface of at Mato Grosso state, Brazil (SIN-01) and (b) for a Gibbsic Geric the kaolinitic clay is low. Ferralsols also have a considerable Ferralsol (Alumic, Hyperdystric, Clayic, Xanthic) at Amapa´ state, capacity to accumulate soil organic matter through prevalent Brazil (JRI-01). organo-mineral interactions and their extensive depth (Dick et al., 2005; Zinn et al., 2007). Although often with a low out the entire profile, the only exception being the surface volumetric water holding capacity (due to their low silt con- layer where concentrations are twice as high, most likely due tent), as a consequence of their favourable physical structure to organic matter content and nutrient recycling. Neverthe- and considerable depth Ferralsols may be capable of storing less, aluminium increases towards the bottom of the profile as much more water than the other more common tropical soil a result of saprolite weathering. The prefix “Acric” refers to types, this allowing forests on such soils in eastern Amazo- the slight increment of clay with increasing depth, which is in nia to maintain physiological activity throughout extended turn clearly dominated by sand. Carbon content is relatively dry seasons (Lloyd et al., 2009; Sect. 3.7) low and mirrors base cation distributions. This suggests that In this study, Ferralsols were found in central and eastern organic matter may be a major source for nutrients and CEC areas of Amazonia as well as towards its southern border. in this soil. No Ferralsols were found in the geologically younger areas The other profile shown is a Gibbsic Geric Ferralsol (Alu- close to the Andes, in agreement with the previous analy- mic, Hyperdystric, Clayic, Xanthic), located in north-eastern sis of Sanchez and Buol (1975). As mentioned already, Fer- Amazonia, Brazil (JRI-01, Fig. 9b), and an example of what ralsols were once thought to dominate the Amazonian land- is often called “Belterra clay” in the soils literature (Som- scape, but this was later shown to be an erroneous perception broek, 1966). As opposed to the Sinop Ferralsol, this soil arising from a low reliability of soil mapping in the area (van has a very high clay proportion (> 0.75), which may be par- Wambeke et al., 1983; Richter and Babbar, 1991). For exam- tially responsible for its relatively high carbon content (Dick ple, Sombroek (1966) suggested that the area covered by Fer- et al., 2005). Although being still at the lower fertility end of ralsols (especially the “Belterra clay” soils), was much larger the spectrum, exchangeable bases and aluminium are almost than we now know to be the case. Indeed, later in his career five times higher than for SIN-01. The qualifiers “Gibbsic” he reported that initial assessments were overestimated, and and “Xanthic” refer to presence of significant amounts of the that the areas covered by Belterra clay in western areas of mineral gibbsite and to the inherent yellow colour common Amazonia were much more limited than he first thought. to this soil, respectively. This diversity in chemical and mor- Two Ferralsol profiles are shown in Fig. 9. The first is phological characteristics demonstrates the high variability a Geric Acric Ferralsol (Alumic, Hyperdystric, Arenic) lo- and diversity even within the most highly weathered RSG cated in Mato Grosso state, Brazil, at the southern fringe groups. of the Amazon forest (SIN-01, Fig. 9a). Consistent with its description as “Geric”, the concentration of exchangeable bases and aluminium in this Ferralsol are very low through- www.biogeosciences.net/8/1415/2011/ Biogeosciences, 8, 1415–1440, 2011 1426 C. A. Quesada et al.: Soils of Amazonia with particular reference to the RAINFOR sites

-1 -1 Exchangeable cations (mmolc kg soil) Carbon concentration (mg g ) Particle size distribution imum in the zone of increased silt content. Such vertical 0 20 40 60 80 0 5 10 15 0.00 0.25 0.50 0.75 1.00 0.0 gradients in the nature of active weathering reflect the inter- 0.2 Clay mediate pedogenetic status of Amazonian Alisols. 0.4 Silt 0.6 Sand The Ecuadorian soil has a slightly different cation and par- Al 0.8 Na ticle size profile. Calcium dominates the top 0.05 m of the 1.0 K Mg 1.2 Ca profile, but below that both [Mg] and [Al] increase with depth Soil depth (m) 1.4 to the point where Mg (and some Ca) dominate the ECEC. 1.6 1.8 (a) Aluminium concentrations reach a maximum in the clay il- 2.0 luviation zone, while [Mg] and [Ca] have their increases

-1 -1 lower down in the profile in association with an increase Exchangeable cations (mmolc kg soil) Carbon concentration (mg g ) Particle size distribution 0 20 40 60 80 0 5 10 15 0.00 0.25 0.50 0.75 1.00 in coarse fraction increments. This again suggests a non- 0.0 Al 0.2 uniform weathering profile, probably in accordance with dif- Na 0.4 K Clay Mg Silt ferent weathering stages for the different soil minerals with 0.6 Ca Sand 0.8 depth. 1.0 Such patterns of [Al] and [Mg] reflect the pedogenetic 1.2 Soil depth (m) 1.4 stage of Alisols as well as the properties of 2:1 clay minerals 1.6 (Driessen et al., 2001). During the early formation of Al- 1.8 (b) 2.0 isols, hydrolysis and transformation of primary weatherable minerals occur in the parent material with some leaching of

Ground water silica. This produces a saprolite with little weatherable pri- mary minerals and a dominance of high activity clays, most Fig. 10 10. Profiles of exchangeable cations, soil carbon and tex- likely formed by the transformation of micas. This is fol- ture: (a) for a Haplic Alisol (Hyperdystric, Siltic) in south Peru lowed by a redistribution of clay in the soil and the forma- (TAM -06) and (a) for a Hyperalic Alisol (Hyperdystric, Clayic) tion of an argic horizon. It is when secondary high activity in Ecuador (TIP-05). clays start to weather that Alisol formation occurs and the ob- served cation patterns become distinguishable. High active clays are unstable in environments that are depleted of silica 3.3.3 Alisols and alkaline and alkaline earth cations. With their weathering they release soluble aluminium and for some parent materi- Alisols (Ultisols in US Soil Taxonomy) are strong acid soils als, iron and magnesium from the octahedral internal layers with high activity clays accumulating in the subsoil. They of the 2:1 clay minerals. The process of high activity clay occur on parent materials which contain a substantial amount weathering usually overlaps with clay redistribution. Clay il- of unstable aluminium-bearing materials such as hydroxy- luviation only occurs at pH varying from 5 to 6.5. At lower interlayered smectite or vermiculite. Ongoing hydrolysis of acidic pH, Al+3 becomes dominant and saturate the complex. these minerals releases Al which then occupies more than Such Al saturation flocculates the clay and impedes further half of the exchangeable sites. Alisol formation is confined dispersion, leading to its accumulation in the profile. to environments where most of the primary minerals have Exchangeable bases in the Alisol profiles were among the disappeared, and with secondary clay minerals of high ac- highest levels found in Amazonian soils. Alisols are gen- tivity dominating the clay complex. They have a CEC in erally considered soils of limited fertility as they have their −1 ammonium acetate pH 7 above 240 mmolc kg clay and a ECEC dominated by aluminium. But the very low fertility base saturation of less than 0.5 in a major part of the profile. levels common to many Amazonian soils make them com- Within the tropics they usually occur on old land surfaces paratively rich, occupying the more fertile end of the spec- with hilly or undulating topography and in this study were trum. Also, as is shown in Quesada et al. (2010), phospho- found in Ecuador, Peru and Colombia. rus supply is also generally favourable in these soils and rea- Two lower level variations of Alisols have their profiles sonably high base cations often co-exist with the high alu- shown here. The first is a Haplic Alisol (Hyperdystric, Siltic) minium levels. Indeed, despite their very high [Al], there in south Peru (TAM-06, Fig. 10a), and the second is a Hy- is little evidence of aluminium toxicity for plants growing peralic Alisol (Hyperdystric, Clayic) in Ecuador (TIP-05, on such soils. For example, Gama and Kiehl (1999) found Fig. 10b). that crops growing in western areas of Amazonia did not Exchangeable [Ca] and [Mg] were relatively high in the shown serious Al toxicity symptoms; even though the soils Peruvian Alisol, along with a high exchangeable [Al]. Al- they were growing on had exchangeable Al levels of approx- −1 though Ca was mostly found near the soil surface, [Mg] were imately 145 mmolc kg . Marques et al. (2002) consider the relatively constant until 1.2 m, increasing steadily with depth. high Al extracted from such soils as most likely originating Most of the Al seems to be being released in the active clay from Al adsorbed within the structure of the 2:1 minerals illuviation zone, but with Mg weathering reaching its max- such as hydroxyl-interlayered smectites. They suggest that in

Biogeosciences, 8, 1415–1440, 2011 www.biogeosciences.net/8/1415/2011/ C. A. Quesada et al.: Soils of Amazonia with particular reference to the RAINFOR sites 1427 situ this “exchangeable” Al may not necessarily be in equi- They are strongly weathered acid soils with low base satura- librium with solution [Al], and perhaps not readily “seen” by tion and acid reaction (Lathwell and Grove, 1986; Richter plant roots. and Babbar, 1991; West et al., 1998; Driessen et al., 2001; The mapped distribution of Alisols (Fig. 4; see also Fig. S5 IUSS Working Group WRB, 2006). However, unlike Al- in the Supplementary Information) suggests only a lim- isols, Acrisols are dominated by low activity clays with a −1 ited coverage of these soils in Amazonia. But this is al- lower CEC range (< 240 mmolc kg clay, extracted by am- most certainly an underestimate. For instance, all Alisols monium acetate pH 7). They usually evolve on acid rock found in this study were found in areas outside of those of Pleistocene age or older, being notably high in strongly identified as such in the SOTERLAC Database. In addi- weathered clays which are still undergoing further degrada- tion, soil profiles found in the RADAMBRASIL soil sur- tion. The sediments that form the parent material for Acrisols vey (RADAMBRASIL, 1978), in particular in the Brazilian often have been pre-weathered for more than one weathering states of Acre and Amazonas (west and southwest part of cycle (West et al., 1998). Nevertheless, geomorphic studies the Brazilian Amazonia), show several soil profiles which in tropical areas have invariably shown that although Acrisols would almost certainly be classified as Alisols in the WRB occupy younger geomorphic positions than Ferralsols, they system. These soils named as podzolicos´ vermelho amarelo also occur in older and more stable areas than those occu- alico´ , have high activity clays, with [Al] ranging from 50 to pied by other soils with which they are often geographically −1 as much as 268 mmolc kg and with other characteristics associated (Beinroth et al., 1974; Lepsch and Buol, 1974; fulfilling all requirements for Alisols. Furthermore, there are Beinroth, 1981; West et al., 1998). Contents of Al, Fe and many other soils in that region having characteristics very Ti oxides are comparable to Ferralsols with their clay frac- similar to that of Alisols, but which most likely have been tion consisting almost entirely of well crystallized kaolinite converted to WRB in the SOTERLAC database as Acrisols. and some gibbsite (Buol et al., 2003). Nevertheless, small This is because in the translation process, such soils were quantities of mica, vermiculite and smectite occur in many generally taken to contain only low activity clays, overlook- of these soils (Sanchez and Buol, 1974; West et al., 1998). ing the fact that the definition of clay activity is very different Also, as for Ferralsols, Acrisols are often reported as having between the WRB and the Brazilian Soil Classification Sys- variable charge clays (Uehara and Gilman, 1981; Qafoku et tem as was used at the time of RADAMBRASIL. Although al., 2004) the WRB classifies on a per unit clay basis, the Brazilian Sys- As reviewed by West et al. (1998) the physical properties tem uses bulk soil CEC values (de Oliveira and van der Berg, of Acrisols (and the related Luvisols, Lixisols and Alisols) 1996). Therefore many soils regarded as low activity clay often present major constraints for plant growth. Textural in the RADAMBRASIL would probably have been consid- variation within the profile is a common problem because of ered as high activity clays soils if CEC had been corrected water perching, limited infiltration and thus increased runoff. for clay content. Another problem regarding the conversion Moreover, the sandy/loamy surface horizons often have weak of clay activity from the Brazilian System is the lack of con- structure which favours compaction and high bulk density. firmation by standard chemical analysis (CEC above or be- These conditions in turn result in low infiltration rates, lim- −1 low 240 mmolc kg of clay, extracted by ammonium acetate ited rooting and seedling emergence. In many of these soils, pH 7 and corrected for carbon content) because such analy- the bulk density is greater than that considered to be limiting sis were generally lacking in the RADAMBRASIL program. for root proliferation due to high mechanical impedance. The As an additional concern the correlation between the Brazil- presence of patches of macro-porosity and localized lower ian and WRB systems is generally weak with this prohibiting bulk density may allow root penetration into horizons of the accurate translation of soil map legends as needed for sci- overall high bulk density, but if roots are restricted to these entific purposes (de Oliveira and van der Berg, 1996). Taken zones the volume of soil exploited for water and nutrients together these observations suggest that many soils described must be limited. as Acrisols in SOTERLAC for the Brazilian Amazon are, in In this study Acrisols were found in north Peru, Venezuela fact, Alisols as defined by the WRB. and Brazil and Fig. 11 shows soil profiles for a Vetic Acrisol (Hyperdystric) located at Mato Grosso state, Brazil (ALF- 3.3.4 Acrisols 01). Exchangeable cation concentrations in the subsoil of this Acrisol are much lower than the surface horizon, and Acrisols (Ultisols in US Soil Taxonomy) are considered the show little vertical variability in sub-soil nutrient concentra- second most common soils in Amazonia, covering an area of tions other than a constant decline. Although soil carbon and approximately 2.15 × 106 km2 (Table 1), although as men- exchangeable cations may show a slight increase (or lower tioned in Sect. 3.3.3, this is almost certainly an overestimate. level of decline) in the clay increment zone, sub-soil concen- As for Alisols, these soils are characterized by accumula- trations can be generalized to be uniformly low throughout tion of clays in a subsurface horizon, which can be a result the profile. This contrasts with other soils within the same of several different processes, including sedimentation, litho- morphological group such as Alisols and Lixisols, where logical discontinuities, and clay migration (West et al., 1998). cation concentrations may increase with depth (Figs. 10 and www.biogeosciences.net/8/1415/2011/ Biogeosciences, 8, 1415–1440, 2011 1428 C. A. Quesada et al.: Soils of Amazonia with particular reference to the RAINFOR sites

-1 -1 Particle size distribution Exchangeable cations (mmolc kg soil) Carbon concentration (mg g ) have base saturation of 0.5 or more. They are considered 0 10 20 30 0.0 2.5 5.0 7.5 10.0 0.00 0.25 0.50 0.75 1.00 0.0 soils with advanced weathering stage but due to their parent 0.2 Al Clay material (often limestone or mafic rocks) base saturation is 0.4 Na Silt K Sand 0.6 Mg still high, with soil reaction eliminating excessive amounts 0.8 Ca of Al in the soil solution (Driessen et al., 2001). 1.0 1.2 Only one Lixisol was identified in this study, a Cutanic Soil depth (m) 1.4 1.6 Lixisol (Ferric, Hypereutric) in Venezuela (RIO-12) and 1.8 Fig. 12 shows profiles for this soil. Exchangeable bases 2.0 clearly dominate the exchange complex, with Mg and Ca occupying most of the ECEC and K, Na and Al present Fig. 11 11. Profiles of exchangeable cations, soil carbon and texture for a Vetic Acrisol (Hyperdystric) at Mato Grosso state, Brazil (ALF- in smaller quantities. Magnesium is in clear excess to Ca 01). which may be a result of parent material chemistry or im- mobilization of Mg during neoformation of secondary min- erals (Thomas, 1974). The concentration of Mg increases in -1 -1 Exchangeable cations (mmolc kg soil) Carbon concentration (mg g ) Particle size distribution 0 10 20 30 0 5 10 15 0.00 0.25 0.50 0.75 1.00 the clay illuviation zone, reaching its highest concentrations 0.0 at the maximum clay content. Below [Al] and [K] increase 0.2 Clay Silt 0.4 slightly with a small reduction in [Mg]. The vertical distribu- Sand 0.6 Al Na tion of elements in this soil suggests that active weathering 0.8 K 1.0 Mg of soil minerals may be still taking place. Although base sat- Ca 1.2 uration is high in Lixisols, the ECEC itself is relatively low. Soil depth (m) 1.4 For the profile examples here, this results in our Hypereutric 1.6 1.8 Lixisol actually having a base cation pool which is actually 2.0 less than the hyperdystric Alisols shown above.

Compact clay layer This Lixisol was also very compact, had unweathered

rocks scattered throughout, and was with a heavy structure in Fig.Fig. 12 12. Profiles of exchangeable cations, soil carbon and texture the B horizon. Although physical constraints are a common for a Cutanic Lixisol (Ferric, Hypereutric) in Venezuela (RIO-12). feature in all soils with increments in clay content, problems such as compaction, low porosity, high bulk density and top soil hardening when dry are particularly common in Lixisols 12, respectively). This suggests that any contribution from (Nicou, 1974, 1975; Nicou and Charreau, 1980; IUSS Work- active weathering of soil minerals may be small for Acrisols, ing Group WRB, 2006). which also implies that such soils are more strongly weath- ered than Alisols and Lixisols. At the soil surface, which is 3.3.6 Nitisols predominantly sand, the higher concentration of base cations presumably arises from the recycling of organic matter. Con- sistent with the “Vetic” qualifier which predicts a cation ex- Nitisols (Ultisols and Oxisols in US Soil Taxonomy) are −1 change capacity below the level of 60 mmolc kg in some deep, red (Rhodic), well-drained tropical soils with a clayey layer within the top 1 m of soil, this Acrisol CEC resulted nitic subsurface horizon that has typical polyhedric, blocky below that level in all parts of the profile. However, despite structure elements with shiny ped faces. Usually they lower cation concentrations than other soils with similar tex- evolve from the weathering of intermediate to basic parent tural gradients, the ALF-01 Acrisol was twice as fertile as the rock, possibly rejuvenated by recent additions of volcanic Amazonian Ferralsol examples. ash. Their mineralogy is dominated by kaolinite and meta- halloysite, but minor amounts of illite, vermiculite and ran- 3.3.5 Lixisols domly interstratified clay minerals may be present along with hematite, goethite and gibbsite. These soils are iron rich soils Lixisols (Alfisols in US Soil Taxonomy) are soils morpho- usually found on level to hilly landscapes (Driessen et al., logically related to Acrisols, Alisols and Luvisols through the 2001). In Brazil, these soils are know as terras roxas es- presence of argic horizons; being separated from Alisols and truturadas and are often highly sought for their agricultural Luvisols by having low activity clays and low CEC. They are capabilities. Nitisols in Amazonia occupy areas where basic separated from Acrisols on the basis of their higher base satu- effusions occur (i.e. diabase and dolerite) such as in outcrops ration. For Lixisols, clay is transported from an illuvial hori- of the Guyana and Brazilian shields and some carboniferous zon to an argic subsurface horizon that has low activity clays deposits (Sombroek, 1966). For example Nitisols were re- and moderate to high base saturation. Similar to Acrisols, ported to occur along the Trans-Amazon highway, in the up- −1 Lixisols have a low CEC range (< 240 mmolc kg clay ex- per Xingu area and in the state of Rondoniaˆ (south western tracted by ammonium acetate pH 7) but unlike Acrisols, they Amazonia), as well as at the Raposa Serra do Sol indigenous

Biogeosciences, 8, 1415–1440, 2011 www.biogeosciences.net/8/1415/2011/ C. A. Quesada et al.: Soils of Amazonia with particular reference to the RAINFOR sites 1429

-1 -1 Exchangeable cations (mmolc kg soil) Carbon concentration (mg g ) Particle size fraction tent in these soils (Richter and Babbar, 1991). At 1.7 m a 0 10 20 30 40 50 0 10 20 30 40 0.00 0.25 0.50 0.75 1.00 0.0 compact clay layer with scattered rocks having vitric proper- Clay 0.2 Al Silt ties constrained sampling to that layer. 0.4 Sand Na 0.6 K 0.8 Mg 3.3.7 Podzols 1.0 Ca 1.2 Soil depth (m) 1.4 Podzols (Spodosols in US Soil Taxonomy) are soils with 1.6 whitish grey subsurface horizon, bleached by organic acids, 1.8 2.0 this overlying a dark accumulation horizon with brown or black illuviated humus. Podzols occur in humid areas, in par- compact clay layer ticular in the boreal and temperate zones but locally also in the tropics. In Amazonia they develop over unconsolidated Fig. 13 13. Profiles of exchangeable cations, soil carbon and texture weathering materials of siliceous rock which are prominent for a Vetic Nitisol (Hypereutric, Rhodic) in Bolivia (HCC-22). on alluvial, colluvial and aeolian deposits of quartzitic sands. They occur mostly along the Rio Negro and in the northern upper Amazon Basin (Do Nascimento et al., 2004). reserve in northeast Roraima (Melo et al., 2010), but with a During the formation of Podzols, complexes of Al, Fe and typically limited and patchy coverage (Sombroek, 1984). organic compounds migrate from the surface soil to the B The formation of Nitisols involves an initial process of fer- horizon with percolating rainwater. Podzolisation is a com- ralisation, similar to that for Ferralsols but with this still be- bination of processes, including the movement of soluble ing at an early stage. The formation of Nitic properties fol- metal-humus complexes (chelates) out of the surface soil to lows with strong angular, shiny peds being developed in the greater depth (cheluviation), and the subsequent accumula- subsurface horizon. This nitidisation is thought to be a re- tion of Al and Fe chelates in a spodic horizon (“chilluvia- sult of alternating micro swelling and shrinking and produces tion”; Driessen et al., 2001). There are contrasting views well-defined structural elements with strong, shiny pressure regarding the formation of Podzols in the tropics, most likely faces. These soils also involve a strong biological influence, with one or another occurring in different environmental con- termite, ants and worm activity is thought to homogenize and ditions. Some authors advocate that Ferralsols and Acrisols create further structure and gradual diffuse horizon bound- (or eroded material from them) can undergo transformations aries (Driessen et al., 2001). The cation exchange capacity and form Podzols under water saturation, via selective clay of Nitisols is high compared to that of similar tropical soils removal and lateral movement processes (Lucas et al., 1984; such as Ferralsols, Lixisols and Acrisols. The reasons for Chauvel et al., 1987; Bravard and Rihgi, 1989; Lucas, 1997; this higher fertility lies in their characteristically high clay Dubroeucq and Volkoff, 1998; Do Nascimento et al., 2004). content and relatively high levels of organic matter. Base But Podzols can also evolve locally as a consequence of ver- saturation varies from 0.1 to 0.9 in such soils (Richter and tical pedogenic processes as described by Horbe et al. (2004). Babbar, 1991). Despite that, Nitisols also show clay incre- Despite their relatively small area (Table 1) Podzols are one ments with depth, although usually they have slightly better of the most studied soil types in Amazonia (Vitousek and structure than those other soils with similar properties. This Sanford, 1986), with a great variety of studies investigating can be taken as a demonstration of the beneficial effect of their characteristic infertility and associated tight nutrient cy- organic matter and soil organisms on soil structure. cling (Went and Stark, 1968; Jordan and Herrera, 1981; Jor- Figure 13 shows soil profiles for a Vetic Nitisol (Hypereu- dan, 1989). A comprehensive review of formation and char- tric, Rhodic) in Bolivia (HCC-22), the only Nitisol identified acteristic of Amazonian Podzols is given by Do Nascimento in this study. Cation profiles show that the exchange capacity et al. (2004). of this soil is high, with [Ca] and [Mg] accounting for most Generally these soils have severe acidity, high [Al], low of it. The vertical distribution of cations shows a gradual de- chemical fertility and unfavourable physical properties. The crease with depth, with [Mg] remaining constant throughout organic matter profile of Podzols usually shows two areas the soil profile. At depths greater than 1.4 m [Mg] becomes of concentration, one at the surface and one in the spodic the dominant cation in the profile, with other cations appear- horizon, but often erosion of surface horizons can expose the ing just in trace amounts. Clay increases considerably in the albic horizon (a low organic matter, bleached layer) result- subsurface but then experiences a gradual decline with silt ing in only one zone of increased organic matter. Usually, contents increasing substantially with depth. Soil carbon is high concentrations of Al or Fe occur in conjunction with the generally high but remarkably so in the soil surface, show- spodic horizon, this reflecting the concentrations of metal- ing only a small decline with depth. Vertical distribution of humus chelates in this zone. The C:N-ratio is typically as carbon, cations and particle size fraction are characteristic of high as 50 in the surface horizon, and nutrient levels in Pod- Nitisols, where high cation exchange capacity and high base zols are low as a consequence of a high degree of leaching saturation is often associated with high organic matter con- (Buol et al., 2003; Quesada et al., 2010). Plant nutrients are www.biogeosciences.net/8/1415/2011/ Biogeosciences, 8, 1415–1440, 2011 1430 C. A. Quesada et al.: Soils of Amazonia with particular reference to the RAINFOR sites

-1 -1 -1 -1 Exchangeable cations (mmolc kg soil) Carbon concentration (mg g ) Particle size distribution Exchangeable cations (mmolc kg soil) Carbon concentration (mg g ) Particle size distribution 0 2 4 6 0 10 20 30 40 0.00 0.25 0.50 0.75 1.00 0 50 100 0 20 40 60 0.00 0.25 0.50 0.75 1.00 0.0 0.0

0.2 0.2 Clay Al Clay Silt Na Silt 0.4 0.4 Sand K Sand 0.6 Mg 0.6 Ca 0.8 0.8 Al -1 1.0 23.9 mmolc kg 1.0 Na K 1.2 1.2 Mg Soil depth (m) Soil depth (m) 1.4 1.4 Ca 1.6 1.6 1.8 1.8 2.0 2.0

Continuation of orstenic layer Rocks to unknown depth Fig. 15 Fig. 14 14. Profiles of exchangeable cations, soil carbon and texture Fig. 15. Profiles of exchangeable cations, soil carbon and texture for a Ortsteinc Podzol (oxyaquic) in Colombia (ZAR-01). for a Haplic Fluvisol (Orthodystric) in Ecuador (JAS-05). concentrated in the surface horizon where impressive root 3.4 Soils conditioned by topography and drainage mats often occur (Jordan and Herrera, 1981). Nutrient econ- 3.4.1 Fluvisols omy in these soils is highly dependent on the recycling of elements which are released by decomposing organic debris. Fluvisols (Entisols–Fluvents in US Soil Taxonomy) are ge- In hydromorphic Podzols, dissolved organic matter bound netically young soils formed on sediments of alluvial ori- with Al is often transported laterally as shallow groundwaters gin (Buol et al., 2003). Their principal morphologic char- restrict vertical transport within the soil. These hydromor- acteristic is the presence of stratification and weak horizon phic Podzols with lateral water flow are inevitably associ- differentiation (Driessen et al., 2001; IUSS Working Group ated with “black water” rivers and lakes in boreal, temperate WRB, 2006). They are usually shallow with their fertility and tropical areas (Driessen et al., 2001). The formation of strongly dependent on the nature of the material deposited a hardpan by illuviated sesquioxides and organic matter (ort- (Irion, 1984; Sanchez, 1976). The only Fluvisol identified in steinic) is common in podzols where there is periodic water this study (JAS-05) had been derived from lateral movements stagnation in the soil, either in the B horizon or below it. of the Napo River in Ecuador. Fluvisols are a common soil Water movement through the soil may be restricted, even in group in Amazonia, especially in its western parts where they upland areas, if such a dense illuviated horizon or an indurate occur over large floodplains of recent alluvial origin (Som- layer is present (Buol et al., 2003). broek, 1984; Fig. 4). In this study, Podzols were found in Brazil, Colombia Soil profile data for our Haplic Fluvisol (Orthodystric), is and Venezuela, usually supporting forests, with a low above shown in Fig. 15. Exchangeable bases were relatively high ground biomass. Together with the Arenosols (Sect. 3.5.1) but with low [Al]. Despite such high nutrient concentrations, they are the most infertile soils in Amazonia and Fig. 14 this soil was still classified as Orthodystric as its base satu- shows profiles for one Ortsteinc Podzol (oxyaquic) in ration extracted by ammonium acetate pH 7 was only 0.21, Colombia (ZAR-01). The vegetation over this soil is locally a problem already discussed in Sect.2.2. The carbon profile called Varrillal which translates to “land of twigs”; it is a showed a relatively high [C] near the surface, followed by stunted forest in which thin and very short trees are abun- a decline until 0.4 m, rising again at depth, this also being dant. An impressive root mat covers the soil, almost no roots associated with irregular variations with depth for the parti- were observed inside the mineral soil itself. The exchange- cle size distributions. These stratification patterns probably able bases are very low with most found in the very topsoil. reflect the discrete sediment deposition events which are typ- Aluminium constitutes most of the ECEC and an association ical for this soil type. The lower portion of the soil profile between base cation concentrations and soil organic matter is consisted of rounded rocks similar to that found at the mar- obvious; most if not all nutrient exchange must occur in the gins of the Napo River itself, suggesting that the area occu- root mat which keeps the remaining nutrients held within the pied by today’s soil was once under or at the river margin. system. At the start of the ortsteinic layer at approximately Above that, two different depositional layers were clearly 1 m deep, aluminium concentration increased abruptly as did identifiable, with the lower buried soil layer having a rela- the carbon content. This is taken to reflect the composition of tively higher [C] than the layer immediately above. the ortsteinic layer where Al-humus chelates act as cement- ing substances.

Biogeosciences, 8, 1415–1440, 2011 www.biogeosciences.net/8/1415/2011/ C. A. Quesada et al.: Soils of Amazonia with particular reference to the RAINFOR sites 1431

-1 -1 Exchangeable cations (mmolc kg soil) Carbon concentration (mg g ) Particle size distribution are shown here, these representing the two ends of the fertil- 0 50 100 0 10 20 30 40 0.00 0.25 0.50 0.75 1.00 0.0 ity spectrum. Figure 16a is for Haplic Gleysol (Orthoeutric,

0.2 Al Clay Siltic), located in Ecuador (TIP-03). For this soil, exchange- Silt 0.4 Na K Sand able bases were very high throughout the profile, at least by 0.6 Mg 0.8 Ca Amazonian standards, with [Ca] being particularly high, and 1.0 Al saturation low. Carbon content was, however, only promi- 1.2 Soil depth (m) 1.4 nent at the very surface of the soil in the layer with some soil 1.6 structure apparent. Below that depth, C contents are low and 1.8 (a) constant throughout the profile. Particle size fractions sug- 2.0 gested a silty clay, changing to clay in some soil layers. Water Table The second profile is for a Haplic Gleysol (Alumic, Hy- -1 -1 Exchangeable cations (mmolc kg soil) Carbon concentration (mg g ) Particle size distribution 0 10 20 30 40 0 10 20 30 40 0.00 0.25 0.50 0.75 1.00 perdystric) in Colombia (ZAR-02, Fig. 16b). Unlike TIP-03, 0.0 Al exchangeable bases were low but with the CEC quite high 0.2 Clay Na Silt 0.4 K and dominated by [Al]. Despite surface soil C being not as Sand 0.6 Mg Ca 0.8 high as in the first Gleysol example, the overall distribution 1.0 pattern of C is similar in both soils. Nevertheless, in contrast 1.2 Soil depth (m) 1.4 to TIP-03, the clay fraction was very small with silt and sand 1.6 predominating. Soil particle size was constant throughout 1.8 (b) 2.0 the profile, having a sandy loam texture. From the above it is evident that, as for Cambisols

Water table (Sect. 3.1.2), Gleysols can encompass soils of a wide range of fertilities and textures and with incipient weathering due Fig. 16 16. Profiles of exchangeable cations, soil carbon and texture to perching water often resulting in little differentiation for (a) a Haplic Gleysol (Orthoeutric, Siltic) in Ecuador (TIP-03) throughout the profile. For most Gleysols physical proper- and (b) for a Haplic Gleysol (Alumic, Hyperdystric) in Colombia ties are restrictive; both by water saturation and incipient de- (ZAR-02). velopment of subsurface horizons. Although such soils are usually deep, effective rooting depths are often limited, this being due to constraints imposed by a high bulk density, lim- 3.4.2 Gleysols ited oxygen supplies, soil compactness and water saturation.

Gleysols (Entisols–Aquents in US Soil Taxonomy) are soils 3.4.3 Umbrisols from wetlands which remain saturated for long enough peri- ods to allow formation of gleyic colour patterns; these con- Umbrisols (Umbric Great Groups in US Soil Taxonomy) are sidered to be evidence of reduction processes with or with- a reference group of young soils thought to be restricted to out segregation of iron (Driessen et al., 2001; IUSS Working high altitudes, and accordingly are not even mapped in the Group WRB, 2006). They usually occupy lower positions SOTERLAC database as occurring in Amazonia. The main in the terrain and are linked to shallow groundwater. Such characteristic of these soils is the development of an umbric soils are formed under excessive wetness at shallow depths surface horizon, without any other mature diagnostic sub- for some period of the year or throughout the year. Low re- surface horizon. Organic matter is thought to accumulate in dox conditions brought about by prolonged soil saturation in such soils as a result of low temperatures. They may share the presence of dissolved organic matter induce the reduction some characteristics with Cambisols but a higher position in of ferric iron to mobile ferrous compounds (Osher and Buol, the classification key gives them prevalence. An umbric hori- 1998). When iron compounds are mobilized and removed, zon is a thick, dark coloured surface horizon with low base the soil material shows its own true colours, which normally saturation and rich in organic matter. No other diagnostic have a low hue. This is the reason why permanently saturated horizon may be present apart of an anthropedogenic, an albic gleyic subsoil layers have neutral, whitish/greyish or bluish or a cambic horizon (Driessen et al., 2001). colours (Driessen et al., 2001; Buol et al., 2003). Gleysols Although at odds with the idea of having an umbric layer occurring in land depressions or at the bottom of slopes are arising only as a result of lower temperatures in mountainous usually comparatively fertile compared to adjacent soils as regions (IUSS Working Group WRB, 2006), two adjacent they tend to have a finer texture, slower organic matter de- profiles in the lowland forests of Bolivia were classified in composition rates and an alluvial or colluvial influx of nutri- this work as Umbrisols and Fig. 17 shows profiles for LSL- ents. 01, an Endogleyic Umbrisol (Alumic, Hyperdystric). Fol- Gleysols are common soils in Amazonia (Sombroek, lowing Umbrisol classification requirements, dark organic 1966, 1984; Sanchez, 1976; Fig. 4) and in this study they matter dominates the surface soil. Exchangeable bases are were found in Peru, Ecuador and Colombia. Two soil profiles low, with the CEC dominated by Al with the particle size www.biogeosciences.net/8/1415/2011/ Biogeosciences, 8, 1415–1440, 2011 1432 C. A. Quesada et al.: Soils of Amazonia with particular reference to the RAINFOR sites

-1 -1 -1 -1 Exchangeable cations (mmolc kg soil) Carbon concentration (mg g ) Particle size distribution Exchangeable cations (mmolc kg soil) Carbon concentration (mg g ) Particle size distribution 0 10 20 0 25 50 0.00 0.25 0.50 0.75 1.00 0 1 2 3 0 5 10 15 0.00 0.25 0.50 0.75 1.00 0.0 0.0 0.2 Clay 0.2 0.4 Silt 0.4 Al Clay Al Sand Na Silt 0.6 0.6 Na K Sand 0.8 K 0.8 Mg Mg 1.0 1.0 Ca Ca 1.2 1.2 Soil depth (m) Soil depth (m) 1.4 1.4 1.6 1.6 1.8 1.8 2.0 2.0

Water Table Fig. 18 18. Profiles of exchangeable cations, soil carbon and texture for a Haplic Arenosol (Hyperdystric) in north Peru (ALP-21). Fig. 17 17. Profiles of exchangeable cations, soil carbon and texture for a Endogleyic Umbrisol (Alumic, Hyperdystric) in Bolivia (LSL- 01). ity, low specific heat, and often minimum nutrient contents (van Wambeke, 1992). They also have a limited potential to profile showing a dominance of clay loam. Both clay and supply nutrients through weathering and thus almost all nu- sand showed some increment starting at 0.5 m but it is im- trient supply must come from mineralization of organic mat- possible to tell where they extended as our sampling stops at ter and/or atmospheric deposition. In a recent study, Harter- 0.7 m due to groundwater being reached. mink and Huting (2008) showed that cation exchange capac- ity in 150 Arenosols of Southern Africa varied markedly with Although misclassification cannot be ruled out because small increments in clay content, CEC ranged from about our sampling profile was limited in depth by groundwater, 10.0 to 90.0 mmol kg−1, varying linearly along a change in the WRB soil classification system led unequivocally to us c clay content of only 0.12. A similar relationship was also keying these soils out as such. If not Umbrisols, these soils found in soil carbon content which varied from about 0.5 in Bolivia would probably be classified as Cambisols or per- to 12 g kg−1 for a similar change in clay. Carbon concen- haps Gleysols. A similar situation is reported by Schad et trations and CEC were also linearly related suggesting that al. (2001), who classified several soils as Phaeozems in low- organic matter is the major source of nutrients. However, the land Bolivia. Phaeozems are fertile soils with mollic or um- African Arenosols studied seem to be much more fertile than bric surface horizons, which had been thought to be exclu- those we have observed in the Amazon Basin. For exam- sively associated to the steppes of temperate regions. As a ple, the CECs for both Arenosols found in this study were new group in the WRB system, it seems likely that some less than 4.0 mmol kg−1. Although these soils often support Umbrisols apparently outside the central concept and char- c savannas or grasslands in the tropics, they are also found un- acteristics will appear. It is thus likely that some additional der short forests and even under lowland tropical forests in characteristics will be required to fully separate these soils Amazonia, (Sanchez, 1976; Sombroek, 1966, 2000). Some- into meaningful sub-categories. times, Arenosols are in fact giant Podzols which key out as 3.5 Soils conditioned by parent material Arenosols because the diagnostic spodic horizon is too deep to allow their classification as Podzols. In places where evap- 3.5.1 Arenosols orative demand exceeds water supply, water deficit is con- sidered a major problem for vegetation (Sanchez, 1976), al- Arenosols (Entisols-Psamments in US Soil Taxonomy) are though roots are often found in these soils extending to con- poorly studied soils, most likely a consequence of their lim- siderable depths. ited agricultural importance. This is despite their consid- Figure 18 show soil profiles for a Haplic Arenosol (Hyper- erable worldwide extent, covering 900 million ha globally dystric), a typical “white sand” of northern Peru. Exchange- compared with 700 million ha of Ferralsols (Hartermink and able bases and aluminium are extremely low with most of the Huting, 2008). In Amazonia, their coverage area is much cations pool present at the very upper soil layers, most likely lower (0.03 of the area, approximately 20 million ha, Ta- reflecting nutrients held on soil organic matter. Fertility of ble 1). Arenosols are characterized by coarse textures which such soils is extremely dependant on biological cycles which must be derived from sand rich parent materials transported maintain nutrients held in the biomass (Sombroek, 1966) by wind, water or in some cases, locally weathered and de- with soil carbon concentration also very low and declining posited in colluvial zones through selective erosion (Driessen dramatically with depth. The low carbon and cations con- et al., 2001; Buol et al., 2003; IUSS Working Group WRB, centrations both reflect the characteristic coarse soil texture 2006). Arenosols are generally weakly developed, with lit- of Arenosols. tle horizon differentiation. Characteristic properties of such soils are high water permeability, low water holding capac-

Biogeosciences, 8, 1415–1440, 2011 www.biogeosciences.net/8/1415/2011/ C. A. Quesada et al.: Soils of Amazonia with particular reference to the RAINFOR sites 1433

-1 -1 3.5.2 Andosols Exchangeable cations (mmolc kg soil) Carbon concentration (mg g ) Particle size distribution 0 20 40 60 80 0 100 200 0.00 0.25 0.50 0.75 1.00 0.0 Andosols (Andisols in US Soil Taxonomy) are dark soils de- 0.2 0.4 veloped on volcanic materials. They mostly occur in undu- Al 0.6 Na K 0.8 lating to mountainous zones in humid, arctic to tropical re- Mg Clay 1.0 Ca Silt gions and under a wide range of vegetation types. They are Sand 1.2 Soil depth (m) formed through rapid weathering of porous volcanic mate- 1.4 Sampling ceased at rial which results in accumulation of stable organo-mineral 1.6 C horizon 1.8 complexes and short range order minerals such as allophane, 2.0 imogolite and ferrihydrite (Richter and Babbar, 1991; Shoji Sampling ceased at et al., 1993; Driessen et al. 2001). Their coverage in Ama- C horizon zonia is small and mostly restricted to mountain regions in the Andes, usually occurring under cloud forests. Therefore, Fig. 19. Profiles of exchangeable cations, soil carbon and texture for aFig. Silandic 19 Andosol (Hyperdystric, Siltic) near the Sumaco volcano, Andosols should not be thought of as common soils for low- Ecuador (SUM-06). land tropical vegetation. Following the description in Driessen et al. (2001), these soils are characterised by the presence of either an andic hori- low weathering levels and with the subsurface horizons still zon or a vitric horizon. An andic horizon is rich in allo- lacking development. phanes and similar minerals, or aluminium humus complexes whereas a vitric horizon contains volcanic glass. Andosol 3.6 Other soils formation depends essentially on rapid chemical weathering of porous, permeable, fine-grained mineral material in the A few soil groups known to occur in the humid tropics were presence of organic matter. The high [Al] protects the or- not represented in this study, and because of their limited ganic component of the Al humus complex against biological coverage are not discussed here. These include the Solon- decay. The mobility of these complexes is limited because chaks, Solonetzs, Phaeozems, Regosols, Histosols and Luvi- rapid weathering yields sufficient Al and Fe to produce com- sols. The first two soil groups are unlikely to be found un- plexes with a high metal/organic ratio which have moderate der forest vegetation in Amazonia as Solonchaks are imper- solubility. Together, the low mobility and high resistance to fectly drained, salt rich soils typical of coastal regions with biological degradation promotes an accumulation of organic the Solonetzs being another Na-rich soil group. Phaeozems matter in the topsoil which culminates in the formation of are fertile, wet soils characterised by an extensive accumu- a melanic surface horizon with high organic matter content. lation of organic matter and again with only a very limited The average organic matter content of the surface horizon cover in Amazonia. is about 80 mg g−1 but the darkest profiles may contain as On the other hand, Regosols, Histosols and Luvisols prob- much as 300 mg g−1 organic matter. Soil bulk density is usu- ably have a wider distribution beneath Amazonian vege- ally low in such soil, a consequence of increments in the pore tation. Regosols are classified as suborders Orthents and fraction as weathering advances. The dominant type of clay Psamments in the Soil Taxonomy and are thin soils derived changes over time, particularly in the subsoil, as allophane from unconsolidated material and lacking diagnostic hori- and imogolite are transformed to other clay minerals such as zons. They are thus soils with incipient development. His- halloysite, kaolinite or gibbsite. Aluminium from the humus tosols are soils with > 0.30 organic matter at the surface, complexes also gradually becomes available, with ferrihy- usually occurring in Amazonia only in landscape depres- drite eventually turning into goethite. Eventually, Andosols sions and wetlands, mostly appearing in small patches as- may evolve into a more mature soil, such as a podzol, or a sociated with other soils. Luvisols are similar to Alfisols in soil with ferric properties or clay illuviation (Driessen et al., the Soil Taxonomy system; their principal characteristics are 2001). clay illuviation and both a high ECEC and a high base sat- Only one Andosol was identified in this study, a Silandic uration. Despite their limited range in the moist tropics of Andosol (Hyperdystric, Siltic), under a sub-montane for- South America this makes them very valuable for agriculture est near the Sumaco volcano, Ecuador (SUM-06). For this and also for ecological studies (Richter and Babbar, 1991; soil, exchangeable cations were relatively high at the sur- Sanchez and Buol, 1975). face (Fig. 19) declining dramatically with depth. Soil car- Man made soils have not been included here as this work bon reached nearly 200 mg g−1 at the very topsoil in associ- intended to review only undisturbed forest soils in Amazo- ation with exchangeable Al, possibly indicating dominance nia. There is however evidence that such soils often occur of Al humus complexes at that depth. Nevertheless, the or- under forest vegetation in localities where ancient human ganic matter concentration was reasonably high throughout communities have existed (Posey and Balick, 2006). Of the the profile. The finer fractions of the soil decreased almost man made soils, the Amazonian black earths (Anthrosols or steadily with depth, most likely reflecting the influence of Terra Preta do Indio) have received the most attention due www.biogeosciences.net/8/1415/2011/ Biogeosciences, 8, 1415–1440, 2011 1434 C. A. Quesada et al.: Soils of Amazonia with particular reference to the RAINFOR sites

Multiple possibilities regime (Fritsch et al., 2006). Nevertheless, provided that suf- Rock Leptosols Cambisols ficient changes in drainage occur, Gleysols probably should

Fluvisols evolve into more mature soils, most likely involving a Cam- Sediments Regosols Umbrisols AlisolsPlinthosols Lixisols Acrisols Ferralsols Podzols bisol stage, followed by the development of argic or plinthic Gleysols horizons. Fluvisols may also develop to Cambisols once they Nitisols

Conditioning are no longer exposed to new depositional events. However, Andosols Arenosols parent material changes in drainage may also occur increasing water satu-

Pedogenic development ration in the profile which could result in transformation of

iron to ferrous compounds and formation of Gleysols. Also, Fig. 20 some types of parent material can lead to specific soil forms. Fig. 20. Simplified scheme for soil development in Amazonia. For instance volcanic material almost invariably leads to the formation of Andosols. The only Andosol found in this study to their high non-labile soil carbon contents. A comprehen- had the characteristics of a poorly developed soil, thus it was sive description of Amazonian black earths can be obtained assumed to have a transient condition equivalent to an inter- in Lehmann et al. (2003) and Glaser and Woods (2004). mediary position close to Cambisols. Cambisols are soils showing early signs of horizon differ- 3.7 Linking WRB reference soils groups to a general entiation usually occurring on steep hill slopes in Amazonia, scheme of soil genesis but unless erosion keeps pace with subsoil weathering, they will evolve and give rise to more mature soil groups (Buol et From the above analysis, it is clear that the soils of Amazo- al., 2003). The most important factors influencing Cambisol nia encompass a considerable diversity with observed vari- transformations are topography position and parent material; ations in physical and chemical properties strongly associ- this then leading to a large variation in the relative magni- ated with state of pedogenic development (Quesada et al., tudes of the many processes and mineralogical reactions giv- 2010). To contextualize this diversity in a soil evolutionary ing rise to distinct routes of soil formation. Cambisols are frame, a scheme of pedogenic development is thus proposed thus soils in a transitional state, occurring prior to the devel- (Fig. 20). This ordination of the soils was made on the basis opment of any spodic, ferralitic or argic horizons (Buol et al., of their genetic, chemical and morphological characteristics, 2003). typical mineralogy and interpretation of the effects of local Umbrisols are typical of cool mountain landscapes where soil forming factors. temperature controls over decomposition allow the formation In the very early stages of soil formation, different parent of thick humus layer on the topsoil (umbric horizon). Usu- materials give rise to what should be the first source of vari- ally there is no other diagnostic horizon present but cambic ability in soils. The first soils to evolve from rocks are thus horizons may be present (Driessen et al., 2001). In this study, considered to be the Leptosols which are very shallow soils, some lowland soils in Amazonia did key out as Umbrisols, as with incipient weathering of parent material. If the weather- this group has priority over Cambisols which would probably ing of the subsurface soil occurs at higher rates than surface be their classification if the latter were higher up the classifi- erosion, then such soils will most certainly reach a Cambisol cation tree. In the pedogenic scheme here Umbrisols are ten- stage (Buol et al., 2003). Nevertheless, many soils in Ama- tatively placed after Cambisols, this placement considering zonia develop over deposited sediments and if drainage con- their incipient subsurface development and the development ditions are good, such sediments would most likely evolve to of umbric horizon. a Regosol stage, equivalent to Leptosols but of a sedimentary The intermediate – mature phase of soil development in- origin. After that, different paths of soil formation are pos- cludes some soils with similar weathering levels which make tulated to occur depending on drainage extent. If sediments separation more difficult than for the earlier stages. Some are water saturated for a long time Gleysols may be formed. of these soils share great morphological similarities and thus These can thus be considered to be genetically young soils, their differentiation was based on mineralogy and clay activ- with very little profile development and with this condition ity, cation exchange capacity and characteristics of nutrient being maintained by continuous or periodic water satura- release in vertical soil profiles, this being considered together tion. But for sediments deposited sequentially and where with soil morphology. Alisols are high CEC soils in which this has been in the not too distant past, then Fluvisols oc- high activity clays accumulate in a subsurface horizon. Their cur. Although Fluvisol drainage conditions are usually better mineralogy is dominated by Al bearing secondary clay min- than Gleysols, seasonal saturation of the subsoil still main- erals such as smectite and vermiculite. Such minerals are tain such soils at low level of pedogenic development. formed from the alteration of micas, most commonly found Moisture regime and landscape position are possibly the in Cambisols, suggesting Alisols have some developmental most important factors determining soil formation at these proximity to this soil type. Also active weathering of 2:1 early stages, and development towards other soil types may high activity clay minerals most likely releases aluminium demand some drastic changes in the landscape and water and some base cations to the soil solution, this varying with

Biogeosciences, 8, 1415–1440, 2011 www.biogeosciences.net/8/1415/2011/ C. A. Quesada et al.: Soils of Amazonia with particular reference to the RAINFOR sites 1435 78

soil depth. Although the presence of clay illuviation places 1 - Alisols at a higher evolutionary stage than the previously dis- 1,000 c cussed soils, their higher CEC, high activity clay mineralogy and vertical patterns of weathering/nutrient release suggests 100 that Alisols have a lower level of pedogenic development than the other intermediate soil groups. 10 Plinthosols follow a somewhat different route of forma- tion, being strongly influenced by fluctuation of groundwa- 1 ter, most likely evolving from Gleysols, Fluvisols and some Total Reserve Bases (mmol kg ) Cambisols with aquic properties. They are placed after Al- isols due to their lower CEC but still have similar chemical characteristics, a placement which is also supported by their Leptosols Gleysols 1 Gleysols 2 Fluvisols Andosols Cambisols 1 Cambisols 2 Cambisols 3 Umbrisols Alisols 1 Alisols 2 Alisols 3 Plinthosols 1 Plinthosols 2 Nitisols Lixisols Acrisols 1 Acrisols 2 Ferralsols 1 Ferralsols 2 Podzols Arenosols mineralogy which tends towards the formation of kaolinite Reference Soil Group and sesquioxides. Plinthosols are tentatively separated here from Nitisols, Lixisols and Acrisols, by the ongoing removal of silica and bases and subsequent segregation of iron. RSG Subgroup Prefixes Suffixes Nitisols can be placed between Plinthosols and the remain- Gleysols 1 Haplic Hyperdystric Alumic Gleysols 2 Haplic Orthoeutric ing soils with argic horizon due to their mineralogy and inter- Cambisols 1 Plinthic, Vertic or Haplic Orthoeutric mediate age morphologic characteristics and, compared with Cambisols 2 Stagnic or Vetic Hyperdystric Lixisols and Acrisols, differentiated on the basis of weather- Cambisols 3 Haplic Hyperdystric Alumic ing of parent material. This still seems to be supplying ade- Alisols 1 Haplic Hyperdystric quate amounts of exchangeable bases for Lixisols, but with Alisols 2 Hyperalic Hyperdystric Acrisols not showing any sign of active mineral weathering Alisols 3 Haplic Hyperdystric Alumic Plinthosols 1 Endostagnic Hyperdystric Alumic in their subsoil. Plinthosols 2 Haplic Hyperdystric Alumic; Orthodystric Alumic Acrisols are, however, easily separated from Ferralsols on Acrisols 1 Vetic Hyperdystric the basis of their contrasting morphologies, mineralogy and Acrisols 2 Vetic Hyperdystric Alumic chemistry. For example, unlike Acrisols, Ferralsols do not Ferralsols 1 Geric Hyperdystric Alumic have significant signs of clay illuviation, and have reached Ferralsols 2 Geric Acric Hyperdystric; Hyperdystric Alumic complete desilication and transformation of clay minerals to- ward kaolinite and Fe and Al oxides. Ferralsols might be Fig. 21. Relationship between the axis of soil development and the final stage of weathering in many conditions, but further theFig. chemically 21 based weathering index Total Reserve Bases (6RB). Details for World Reference Base lower level classification are also transformation to Podzols via selective clay removal under water saturation and lateral movement processes has been given. suggested (Lucas et al., 1984; Chauvel et al., 1987; Bravard and Rihgi, 1989; Lucas, 1997; Dubroeucq and Volkoff, Figure 21 shows the relationship between the suggested 1998). Usually, Ferralsols show no tendency to develop an axis of pedogenic development and a chemically based elluvial horizon but they may potentially transform to Pod- weathering index (Total Reserve Bases, 6RB), which is taken zols if iron compounds are removed, and the clay is decom- to be a good proxy for the amount of weatherable mineral re- posed by ferrolysis under conditions of periodic water stag- maining (Delvaux et al., 1989). To account for variations nation. This might occur on colluvial deposits transported within the soil groups, each Reference Soil Group was fur- from Ferralsols down slope. Nevertheless, as already dis- ther divided into subgroup types representing the principal cussed, Podzols can also develop as a consequence of ver- distinctions in the lower taxonomic levels (the table with tical pedogenic processes, following the formation of white Fig. 21 gives taxonomic details for RSG lower classification sands (Horbe et al., 2004). levels). Noting the logarithmic y–axis, this chemical weath- Arenosols usually have depositional origin but can also ering index showed good agreement with the proposed na- be formed under weathering of quartz rich rocks. As con- ture of the soil age gradient. Soils with lower pedogenic de- ditioned by their parent material, they show very little hori- velopment all had high levels of 6RB, with the highest val- zon differentiation and for that reason are often considered ues occurring in Cambisols, Fluvisols and Gleysols. Beyond young soils (IUSS Working Group WRB, 2006). Neverthe- these classes 6RB declines gradually as pedogenic develop- less, many Arenosols in the Amazon Basin seem to derive ment increases, reaching its lowest values in the Podzols. from old reworked sediments originating from the Guyana Such changes along the pedogenic gradient are likely to be shield (Fittkau, et al., 1975). Their soil chemistry is also associated with relative concentration of weatherable miner- characteristic of the “terminal stages” of pedogenesis (Que- als, changes in mineral assemblage through processes such sada et al., 2010) and for those reasons are assumed to belong as desilication, neoformation and kaolinization and changes to older development surfaces. in surface area and charge characteristics associated with www.biogeosciences.net/8/1415/2011/ Biogeosciences, 8, 1415–1440, 2011 79 1436 C. A. Quesada et al.: Soils of Amazonia with particular reference to the RAINFOR sites

changes in the mineral matrix (Uehara and Gilman, 1981). Variations within soil lower levels were also found, and in 400 * a) general, the nomenclature of the lower taxonomic levels was 350 successful in grouping similar soils. Within each soil group, not only do the prefixes for the lower levels characterize in- 300 dividual morphological features but on some occasions they 250 were found to also help to differentiate soils in terms of fertil- 200 ity and weathering level. For instance, Ferralsols named with 150 suffix “Geric Acric” had higher 6 than the (just) “Geric” RB ** ones, as did “Endostagnic” Plinthosols in relation to “Hap- 100 lic Plinthosols”. This was independent of sharing the same 50 fertility suffixes and indicates that the chemical and morpho- 0

logical characteristics described by the WRB suffixes and Maximum measured soil depth (cm) prefixes may reflect meaningful and functional intergrades of *** *** weathering and possibly differences in parent material. How- 200 ever, other subgroups in Alisols and Acrisols appeared to be b) less divergent in relation to weathering with no clear differ- 150 entiation in 6RB. In the World Reference Base a more precise level of fer- tility is usually described by qualifier suffixes which, in the 100 case of this study, included “Orthoeutric”, “Hyperdystric” and “Alumic Hyperdystric”. This was particularly informa- tive for Cambisol group, where a distinct gradient was ob- 50 served among the three lower level soils. Levels of fertility

and weathering in these soils were clearly separated by the Maximum rooting depth (cm) 0 different suffixes, irrespective of the prefix variations associ- ated with morphological characteristics. Alisols Nitisols Also, during the process of soil classification, it was found Lixisols Acrisols Podzols Gleysols Fluvisols Andosols Ferrasols Leptosols Umbrisols Arenosols Cambisols that soils with relatively high levels of fertility were still Plinthosols classified as “Hyperdystric” and with the informative differ- Reference Soil Group entiation indicating exceptionally infertile soils only made through the suffix “Alumic Hyperdystric”, indicating a clear Fig. 22. Changes in soil and root depth in Amazonia following the dominance of aluminium over exchangeable bases. There-*Ferralsolssoil were development often deeper gradient. than 4 m, they(a) aremaximum often report soiled to depth reach anddepths(b) greatermaxi- than 10 m fore, soils with the qualifier suffix “Hyperdystric” (when**Do notmum reflect root actual depth. soil depth as sampling usually stoped at saturated layers ∗ Ferralsols were often deeper than 4 m, they are often reported to reach depths greater not associated with suffix Alumic) can actually be taken as***Maximumthan 10 depth m. measured was 200cm, some roots may be deeper than that ∗∗ constituting relatively fertile soils in the Amazonian context Does not reflect actual soil depth as sampling usually stoped at saturated layers. ∗∗∗ Maximum depth measured was 2.0 m, some roots may be deeper than that. (Quesada et al., 2010). Fig. 22 The diagram in Fig. 20 shows a simplified route for soil pedogenic development. But actual soil formation and evo- structure, dynamics and composition (Quesada et al., 2009). lution is, of course, far more complex. Short cuts in the soil Soil depth (defined here as depth to rock or saprolite con- formation process are known to occur. Moreover, rejuve- tact) is another characteristic associated with pedogenetic nating processes can potentially change the direction of soil stage. In general terms, soils tend to increase in depth fol- development. On other occasions, the saprolite from which lowing weathering, with Leptosols varying from as little as soils evolve might already have been well weathered during 0.2 m deep to more than 4 m in the Ferralsols (Fig. 22a). Nev- ancient times, thus leading to the direct formation of highly ertheless, some variation in this trend is expected to occur weathered soils such as Ferralsols (Buol et al., 2003). Soils for soils of sedimentary versus rock origin. This is because can also themselves become the parent material for other soils forming from sediments tend to form deeper soil lay- soils if dramatic changes in weather or topography occur. As ers, even when occupying lower taxonomy levels. For exam- well, they can be buried, transported and re-deposited, and ple, in young and intermediate age soils derived from crys- even disappear if totally eroded. Nevertheless, irrespective talline rocks, such as those found at the Guyana and Brazil- of development pathway, soils tend to change their character- ian shields (Leptosols, Andosols, Nitisols, Lixisols and some istics through time, and so will ultimately develop to geneti- Cambisols), were not observed to have depths greater than cally mature forms. Clear changes in morphology, chemistry 1.5 m, but soils which had evolved on sediments (Gleysols, and mineralogy will occur during this process and such vari- Alisols, Plinthosols and some Cambisols) had maximum ation ultimately exerts a profound effect over the vegetation depths ranging from 1.5 to 3.5 m (Fig. 22a). In addition, all

Biogeosciences, 8, 1415–1440, 2011 www.biogeosciences.net/8/1415/2011/ C. A. Quesada et al.: Soils of Amazonia with particular reference to the RAINFOR sites 1437 soils with low to intermediate taxonomic levels (i.e. lower Supplementary material related to this than Acrisols), had a soil depth less than 3 m, independent article is available online at: of rock or sedimentary origin. More weathered soils such as http://www.biogeosciences.net/8/1415/2011/ Acrisols and Arenosols have maximum soil depths (MSD) bg-8-1415-2011-supplement.pdf. of about 3.5 m while Ferralsols ranged from about 3.5 m to more than 4 m (that being the maximum depth investigated). Maximum soil depth in Podzols was only 1 m in this study, Acknowledgements. We thank all local collaborators in the RAIN- FOR network for their support during sampling and Tom Bishop but this value should be taken with caution once the sampling (University of Sydney) for providing the S-plus code forming the of these soils was compromised by water saturation and deep basis of our fitting of the depth profile functions using R. This work soil often collapsed during sampling. was supported in part by the European Union Fifth Framework Similarly to maximum soil depth, the maximum depth to Programme as part of CARBONSINK-LBA, part of the European which roots can be observed also increases with pedoge- contribution to the Large Scale Biosphere-Atmosphere Experiment netic development (Fig. 22b). However, for most Amazo- in Amazonia (LBA) and recently by the Gordon and Betty Moore nian soils in this study (particularly the low and intermediate Foundation now funding the RAINFOR network. Shiela Lloyd age soils) root maximum depth was shallower than the max- assisted with manuscript preparation. imum observed soil depth. For instance, MSD in Gleysols was about 1.5 m but their associated maximum root depth Edited by: P. Meir (MRD) was only 0.6 m. Likewise, MSD for the Acrisols was about 3 m but their MRD seems limited to around 1.5 m. Roots deeper than 2 m were only observed in Ferralsols and References some Arenosols, with Ferralsols being well known for their very deep root systems (Jipp et al. 1998; Nepstad et al., Aubert, G. and Tavernier, R.: The Soils of Africa, in: Soils of the 1994). Also, we observed that determinants on root depth Humid Tropics, National Academy of Science, Washington DC, are most commonly factors other than rock or saprolite con- USA, 17–44, 1972. tact. We suggest that root depth in Amazon forests is mostly Beinroth, F. H.: Some considerations on soil classification in gen- controlled by soil structure (i.e. bulk density and aeration) eral, and “soil taxonomy” in particular. Departmental Paper – but with other soil – climate – vegetation interactions such Hawaii Agricultural Experiment Station, 49, 18–25, 1981. Beinroth, F. H.: Some highly weathered soils of Puerto Rico, 1: as tree anchorage and water and nutrient availability also im- morphology, formation and classification, Geoderma, 27, 1–73, portant. 1982. Such observations have important implications for ecosys- Beinroth, F. H., Uehara, G., and Ikawa, H.: Geomorphic relation- tem models since incorrect specifications of rooting depth ships of Oxisols and Ultisols on Kauai, Hawaii, Soil Sci. Soc. may give rise to incorrect simulations of the hydrological cy- Am. J., 38, 128–131, 1974. cle and general fluxes of heat, CO2 and moisture in Ama- Bishop, T. F. A., McBratney, A. B., and Laslett, G. M.: Modelling zonia (Harper et al., 2010). Indeed, although Harper et soil attribute depth functions with equal-area quadratic smooth- al. (2010) showed that model parameterization with greater ing splines, Geoderma, 91, 27–45, 1999. soil depths (10 m) improved model ability to predict ecosys- Bravard, S. and Righi, D.: Geochemical differences in an oxisol- tem function in Amazonia, our results suggests that this may spodosol toposequence of Amazonia, Brazil, Geoderma, 44, 29– not be a reasonable approach for much of the Amazon Basin. 42, 1989. Buol, S. W.: Oxisols, in: Encyclopedia of Soil Science, edited by: Although the assumption of deep soil layers seems correct Lal, R. and Marcel Dekker, New York, 948–951, 2002. for eastern and central portions of Amazonia, particularly un- Buol, S. W. and Eswaran, H.: Oxisols, Adv. Agron., 68, 152–195, der Ferralsols which are notoriously deep rooted soils (Nep- 2000. stad et al. 1994; Jipp et al. 1998), Table 1 suggests that Buol, S. W., Southard, R. J., McCracken, R. J., and McDaniel, P. such soils cover only about 30 % of Amazonia, with the ma- A.: Soil Genesis and Classification, 5 Edn., The Iowa State Uni- jority of soils, especially in Western Amazonia not capable versity Press, Ames, Iowa, 494 pp., 2003. of supporting root development to and/or function at such Buurman, P. and Soepraptohardjo, M.: Oxisols and associated soils depths. In that respect, were not the precipitation in the west- on ultramafic and felsic volcanic rocks in Indonesia, in: Red ern part of the Basin substantially greater in amount and with Soils in Indonesia, edited by: Buurman, P., Agricultural Research much shorter dry seasons than in the east (Malhi and Wright, Reports 889, PUDOC, Wageningen, 93–104, 1980. 2004), then quite likely evergreen forest could not be sus- Chauvel, A., Lucas, Y., and Boulet, R.: On the genesis of the soil mantle of the region of Manaus, Central Amazonia, Brazil, Cell. tained there to any great extent. But rather with some form Mol. Life Sci., 43, 234–241, 1987. of drought–deciduous forest dominating on the shallow but Cotler, H. and Maass, J. M.: Tree management in the northwestern generally fertile soils occurring closer to the Andes; and as Andean Cordillera of Peru, Mt. Res. Dev., 19, 153–160, 1999. also would be anticipated for any future drier Amazon cli- de Mesquita Filho, M. V. and Torrent, J.: Phosphate sorption as re- mate (Lloyd et al. 2009). lated to mineralogy of a hydrosequence of soils from the Cerrado region (Brazil), Geoderma, 58, 107–123, 1993. www.biogeosciences.net/8/1415/2011/ Biogeosciences, 8, 1415–1440, 2011 1438 C. A. Quesada et al.: Soils of Amazonia with particular reference to the RAINFOR sites

De Oliveira, J. B. and van den Berg, M.: Relation between the soil 383–409, 1986. units of the FAO-Unesco soil map of the world legend and the Glaser, B. and Woods, W. I.: Amazonian dark earths: Exploration soil classes used in the Brazilian surveys, Technical Paper 29, in space and time, Springer Verlag, Berlin, 216 pp., 2004. ISRIC, Wageningen, 45 pp., 1996. Harper, A. B., Denning, A. S., Baker, I. T., Branson, M. D., Pri- Delvaux, B., Herbillon, A. J., and Vielvoye, L.: Characterization hodko, L., and Randall, D. A.: Role of deep soil moisture in mod- of a weathering sequence of soils derived from volcanic ash in ulating climate in the Amazon rainforest, Geophys. Res. Lett., Cameroon. Taxonomic, mineralogical and agronomic implica- 37, L05802, doi:10.1029/2009GL042302, 2010. tions, Geoderma, 45, 375–388, 1989. Hartemink, A. E. and Huting, J.: Land cover, extent, and properties Dick, D. P., Goncalves, C. N., Dalmolin, R. S. D., Knicker, H., of Arenosols in Southern Africa, Arid Land Res. Manag., 22, Klamt, E., Kfgel-Knabner, I., Simoes, M. L., and Martin-Neto, 134–147, 2008. L.: Characteristics of soil organic matter of different Brazilian Herbillon, A. J.: Mineralogy of Oxisols and oxic materials, in: Soils Ferralsols under native vegetation as a function of soil depth, with Variable Charge, edited by: Theng, B. K. G., New Zealand Geoderma, 124, 319–333, 2005. Society of Soil Science, Lower Hutt, 109–126, 1980. Dijkshoorn, J. A., Huting, J. R. M., and Tempel, P.: Update of the Herrera, R., Jordan, C. F., Klinge, H., and Medina, E.: Amazon 1:5 million Soil and Terrain Database for Latin America and the ecosystems: Their structure and functioning with particular em- Caribbean (SOTERLAC; version 2.0), Report 2005/01, ISRIC – phasis on nutrients, Interciencia, 3, 223–232, 1978. World Soil Information, Wageningen, 2005. Hoorn, C., Wesselingh, F. P., ter Steege, H., Bermudez, M. A., Do Nascimento, N. R., Bueno, G. T., Fritsch, E., Herbillon, A. Mora, A., Sevink, J., Sanmart´ın, I., Sanchez-Meseguer, A., An- J., Allard, T., Melfi, A. J., Astolfo, R., Boucher, H., and Li, derson, C. L., Figueiredo, J. P., Jaramillo, C., Riff, D., Negri, Y.: Podzolisation as a deferralitization process: a study of an F. R., Hooghiemstra, H., Lundberg, J., Stadler, T., Sarkinen,¨ Acrisol-Podzol sequence derived from Palaeozoic sandstones in T., and Antonelli, A.: Amazonia through time: Andean Uplift, the northern upper Amazon Basin, Eur. J. Soil Sci., 55, 523–538, climate change, landscape evolution, and biodiversity, Science, 2004. 330, 927–931, 2010. Driessen, P., Deckers, J., Spaargaren, O., and Nachtergaele, F.: Lec- Horbe, A. M. C., Horbe, M. A., and Suguio, K.: Tropical spodosols ture notes on the major soils of the world, FAO, Rome, 35–37, in northeastern Amazonas State, Brazil, Geoderma, 119, 55–68, 2001. 2004. Dubroeucq, D. and Volkoff, B.: From Oxisols to Spodosols and Irion, G.: Soil infertility in the Amazonian rain forest, Naturwis- Histosols: evolution of the soil mantles in the Rio Negro basin senschaften, 65, 515–519, 1978. (Amazonia), Catena, 32, 245–280, 1998. Irion, G.: Sedimentation and sediments of Amazonian rivers and FAO: Soil Map of the World. Revised Legend, with corrections, evolution of the Amazonian landscape since Pliocene times, in: Technical Paper 20, Reprint of World Soil Resources Report 60, The Amazon Limnology and Landscape of a Mighty Tropical FAO, Rome, 1988. River and its Basin, edited by: Sioli, H. and Junk, W., Dordecht, Fittkau, E. J.: Esboc¸o de uma divisao˜ ecologicaˆ da regiao˜ 201–214, 1984. amazonica,ˆ Proc. Symp. Biol. Trop. Amaz., Florencia y Leticia, ISRIC (International Soil Reference and Information Centre): 1969, 363–372, 1971. Global and national soils and terrain digital databases (SOTER), Fittkau, E. J., Irmler, U., Junt, W. J., Reis, F., and Schmidt, G. W.: World Soil Resources Report No. 74, FAO, Rome, 1995. Productivity, biomass and population dynamics in Amazonian IUSS (International Union of Soil Science) Working Group WRB: water bodies, in: Tropical ecological systems, Trends in terres- World reference base for soil resources 2006: A framework trial and aquatic and research, edited by: Golley, G. B. and Med- for international classification, correlation and communication, ina, E., Springer, New York, Berlin, 289–311, 1975. World Soil Resources Report 103, FAO, Rome, 2006. Fritsch, E., Herbillon, A. J., Do Nascimento, N. R., Grimaldi, EMBRAPA (Empresa Brasileira de Pesquisa Agropecuaria).´ Mapa M., and Melfi, A. J.: From Plinthic Acrisols to Plinthosols and de solos do Brasil, escala 1:5.000.000. Servic¸o Nacional de Lev- Gleysols: Iron and groundwater dynamics in the tertiary sedi- antamento e Conservac¸ao˜ de Solos, Brasil, 1981. ments of the upper Amazon Basin, Eur. J. Soil Sci., 58, 989– Jipp, P. H., Nepstad, D. C., Cassel, D. K., and de Carvalho, C. 1006, 2006. R.: Deep soil moisture storage and transpiration in forests and Furch, K. and Klinge, H.: Chemical relationships between vegeta- pastures of seasonally-dry Amazonia, Climate Change, 39, 395– tion, soil and water in contrasting inundation areas in Amazonia, 412, 1998. in: Mineral Nutrients in Tropical Forests and Savanna Ecosys- Jordan, C. F.: Are process rates higher in tropical forest ecosys- tems, Blackwell, Oxford, 1989. tems? in: Mineral Nutrients in Tropical Forests and Savanna Gaillardet, J., Dupre, B., Allegre, C., and Negrel, P.: Chemical and Ecosystems, edited by: Proctor, J., Blackwell, Oxford, 1989. physical denudation of the Amazon River Basin, Chem. Geol., Jordan, C. F. and Herrera, R.: Tropical rain forests: are nutrients 142, 141–173, 1997. really critical? Am. Nat., 117, 167–180, 1981. Gama, J. R. N. and Kiehl, J. C.: Aluminium influence on plants Kronberg, B. I., Franco, J. R., Benchimol, R. E., Hazenberg, G., grown in a Brazilian red-yellow podzolic soil, Revista Brasileira Doherty, W. and VanderVoet, A.: Geochemical variations in de Ciencia do Solo, 23, 475–482, 1999. Solimoes formation sediments (Acre basin, Western Amazonia), Gee, G. W. and Bauder, J. W.: Particle-size analysis, in: Meth- Acta Amazonica, 19, 319–333, 1989. ods in Soil Analysis, Part 1. Physical and Mineralogical Meth- Kronberg, B. I., Fralick, P. W., and Benchimol, R. E.: Late qua- ods, American Society of Agronomy and Soil Science Society ternary sedimentation and paleohydrology in the Acre foreland of America, edited by: Klute, A., Madison, Wisconsin, USA, basin SW Amazonia, Basin Res., 10, 311–323, 1998.

Biogeosciences, 8, 1415–1440, 2011 www.biogeosciences.net/8/1415/2011/ C. A. Quesada et al.: Soils of Amazonia with particular reference to the RAINFOR sites 1439

Lathwell, D. J. and Grove, T. L.: Soil-Plant Relationships in the Nicou, R.: Contribution to the study and improvement of the poros- Tropics, Ann. Rev. Ecol. Syst., 17, 1–16, 1986. ity of the sandy and clayed-sandy soils in the dry tropical area, Lehmann, J., Kern, D. C., Glaser, B., and Woods, W. I.: Amazonian Agricultural consequences, Agron. Trop., 29, 1100–1127, 1974. dark earths, origin, properties, management, Kluwer Academic Nicou, R.: The problem of caking with the drying out of sandy and Publishers, Netherlands, 505 pp., 2003. sandy clay soils in the arid tropical zone, Agron. Trop., 30, 325– Lepsch, I. F. and Buol, S. W.: Investigations in an Oxisol-Ultisol 343, 1975. Toposequence in S. Paulo State, Brazil, Soil Sci. Soc. Am. J., 38, Nicou, R. and Charreau, C.: Mechanical impedance to land prepara- 491–496, 1974. tion as a constraint to food production in the tropics (with special Lima, H. N., Mello, J. W. V., Schaefer, C. E. G. R., Ker, J. C., reference to fine sandy soils in West Africa), in: Soil Related and Lima, A. M. N.: Mineralogia e quimica de tres solos de uma Constraints to Food Production in the Tropics, IRRI, Manila, topossequencia da bacia sedimentar do Alto Solimoes, Amazonia Philippines, 371–388, 1980. Ocidental, Rev. Bras. Cienc. Solo, 30, 59–68, 2006. Osher, L. J. and Buol, S. W.: Relationship of soil properties to par- Lips, J. M. and Duivenvoorden, J. F.: Regional patterns of well ent material and landscape position in eastern Madre de Dios, drained upland soil differentiation in the middle Caqueta´ basin Peru, Geoderma, 83, 143–166, 1998. of Colombian Amazonia, Geoderma, 72, 219–257, 1996. Pella, E.: Elemental organic analysis, Part 2, State of the art, Amer- Lloyd J., Goulden, M., Ometto, J. P., Fyllas, N. M., Quesada, C. A., ican Laboratory, 22, 28–32, 1990. and Patino,˜ S.: Ecophysiologuy of forest and savanna vegeta- Pleysier, J. L. and Juo, A. S. R.: A single-extraction method us- tion, in: Amazonia and Climate Change, edited by: Keller, M., ing silver-thiourea for measuring exchangeable cations and effec- Gash, J., and Silva Dias, P., American Geophysical Union, Wash- tive CEC in soils with variable charges, Soil Sci., 129, 205–211, ington DC, 463–484, 2009b. 1980. Lucas, Y.: Biogeoqu´ımica em ambiente equatorial: exemplo dos Posey, D. A. and Balick, M. J.: Human impacts on Amazonia: sistemas Latossolos-Podzois´ da Amazonia,ˆ in: Cong. Intern. The role of traditional ecological knowledge in conservation and Geoq, 4, Anais, 9–12, 1997. development, Columbia University Press, New York. 367 pp., Lucas, Y., Chauvel, A., Boulet, R., Ranzani, G., and Scatolini, 2006. F.: Transic¸ao˜ latossolos-podzois´ sobre a formac¸ao˜ barreiras na Qafoku, N. P., Van Ranst, E., Noble, A., and Baert, G.: Variable regiao˜ de Manaus, Amazonia,ˆ Rev. Bras. Cienc. Solo, 8, 325– Charge Soils: Their Mineralogy, Chemistry and Management, 335, 1984. Adv. Agron., 84, 160–217, 2004. Malhi, Y. and Wright, J.: Spatial patterns and recent trends in Quesada, C. A., Lloyd, J., Schwarz, M., Baker, T. R., Phillips, O. the climate of tropical rainforest regions, Phil. Trans. R. Soc. L., Patino,˜ S., Czimczik, C., Hodnett, M. G., Herrera, R., Ar- Lond. B., 359, 311–329, 2004. neth, A., Lloyd, G., Malhi, Y., Dezzeo, N., Luizao,˜ F. J., Santos, Marbut, C. F. and Manifold, C. B.: The soils of the Amazon basin in A. J. B., Schmerler, J., Arroyo, L., Silveira, M., Priante Filho, relation to their agricultural possibilities, Geogr. Rev., 15, 617– N., Jimenez,´ E. M., Paiva, R., Vieira, I., Neill, D. A., Silva, N., 643, 1926. Penuela,˜ M. C., Monteagudo, A., Vasquez,´ R., Prieto, A., Rudas, Marques, J. J. G. S. M., Teixeira, W. G., Schulze, D. G., and Curi, A., Almeida, S., Higuchi, N., Lezama, A. T., Lopez-Gonz´ alez,´ N.: Mineralogy of soils with unusually high exchangeable Al G., Peacock, J., Fyllas, N. M., Alvarez Davila,´ E., Erwin, T., from the western Amazon Region, Clay Mineralogy, 37, 651– di Fiore, A., Chao, K. J., Honorio, E., Killeen, T., Pena˜ Cruz, 661, 2002. A., Pitman, N., Nu´nez˜ Vargas, P., Salomao,˜ R., Terborgh, J., and Medina, E. and Cuevas, E.: Patterns of nutrient accumulation and Ram´ırez, H.: Regional and large-scale patterns in Amazon forest release in Amazonian forests of the upper Rio Negro Basin, in: structure and function are mediated by variations in soil physi- Mineral Nutrients in Tropical Forests and Savanna Ecosystems, cal and chemical properties, Biogeosciences Discuss., 6, 3993– edited by: Proctor, J., Blackwell, Oxford, 1989. 4057, doi:10.5194/bgd-6-3993-2009, 2009. Melo, V. F., Schaefer, C. E. G. R., and Uchoa,ˆ S. C. P.: Indian Quesada, C. A., Lloyd, J., Schwarz, M., Patino,˜ S., Baker, T. R., Cz- land use in the Raposa-Serra do Sol Reserve, Roraima, Ama- imczik, C., Fyllas, N. M., Martinelli, L., Nardoto, G. B., Schmer- zonia, Brazil: Physical and chemical attributes of a soil catena ler, J., Santos, A. J. B., Hodnett, M. G., Herrera, R., Luizao,˜ developed from mafic rocks under shifting cultivation, Catena, F. J., Arneth, A., Lloyd, G., Dezzeo, N., Hilke, I., Kuhlmann, 80, 95–105, 2010. I., Raessler, M., Brand, W. A., Geilmann, H., Moraes Filho, J. Moreira, F. M. S., Nobrega,´ R. S. A., Jesus, E. C., Ferreira, D. F., O., Carvalho, F. P., Araujo Filho, R. N., Chaves, J. E., Cruz Ju- and Perez,´ D. V.: Differentiation in the fertility of Inceptisols as nior, O. F., Pimentel, T. P., and Paiva, R.: Variations in chemical related to land use in the upper Solimoes˜ river region, western and physical properties of Amazon forest soils in relation to their Amazon, Sci. Total Environ., 408, 349–355, 2009. genesis, Biogeosciences, 7, 1515–1541, doi:10.5194/bg-7-1515- Nepstad, D. C., de Carvalho, C. R., Davidson, E. A., Jipp, P. H., 2010, 2010. Lefebvre, P. A., Negreiros, da Silva, G. H. D., Stone, E. T. A., RADAMBRASIL: Levantamento dos Recursos Naturais, Min- Trumbore, S. E., and Viera, S.: The role of deep roots in the hy- isterio´ de Minas e Energia, Departamento Nacional de Produc¸ao˜ drological and carbon cycles of Amazonian forests and pastures, Mineral, Rio de Janeiro, 1978. Nature, 372, 666–669, doi:10.1038/372666a0, 1994. Reatto, A., Correia, J. R., and Spera, S. T.: Solos do Bioma Cerrado: Nelson, D. W. and Sommers, L. E.: Total carbon and total nitrogen, Aspectos pedologicos, in: Cerrado ambiente e Flora. Planaltina, in: Methods of Soil Analysis: Part 3 – Chemical Methods, SSSA edited by: Sano, S. M. and Almeida, S. P., EMBRAPA CPAC, Book Series No 5, SSSA and ASA, edited by: Sparks, D. L., 47–86, 1998. Madison, WI, 961–1010, 1996. Richter, D. D. and Babbar, L. I.: Soil diversity in the tropics, Adv.

www.biogeosciences.net/8/1415/2011/ Biogeosciences, 8, 1415–1440, 2011 1440 C. A. Quesada et al.: Soils of Amazonia with particular reference to the RAINFOR sites

Ecol. Res., 21, 315–389, 1991. Sombroek, W. G. and Camargo, M. N.: Groundwater laterites and Saatchi S. S., Steinenger, M., Tucker, C. J., Nelson, B., and Simard, ironstone soils in Brazil, with examples from the Amazon region, M.: 1km Land Cover and Wetlands Maps of the Amazon Basin. in: Proceedings of the Second International Seminar on Lateri- Data set available from Oak Ridge National Laboratory Dis- sation Processes, edited by: Melfi, A. J. and Carvalho, A., IAG, tributed Active Archive Center, Oak Ridge, Tennessee, U.S.A.,, Sao Paulo, Brazil, 541–552, 1983. 2008. Thomas, M. F.: Tropical Geomorphology, The McMillan Press, Sanchez, P. A.: Properties and Management of Soils in the Tropics, London, 331 pp, 1974. John Wiley & Sons, New York, 618 pp., 1976. Uehara, G. and Gilman, G.: The Mineralogy, Chemistry, and Sanchez, P. A. and Buol, S. W.: Properties of some soils of the upper Physics of Tropical Soils with Variable Charge Clays, Westview Amazon Basin of Peru, Soil Sci. Soc. Am. J., 38, 117–121, 1974. Press, Boulder, Colorado, 1981. Sanchez, P. A. and Buol, S. W.: Soils of the tropics and the world Van Reeuwijk, L. P.: Procedures for soil analysis. International Soil food crisis, Science, 188, 598–603, 1975. Reference Information Centre, ISRIC, sixth edition, Wagenin- Schad, P., Haussermann, U., Ferber, P., and Rinneberg, K.: gen, Netherlands, 120 pp., 2002. Phaeozems or Ferralsols? Low-activity clay soils with mollic van Wambeke, A.: Properties and potential of soils in the Amazon horizons in Bolivian tropical lowlands, European Soil Bureau Basin, Interciencia, 43, 233–241, 1978. Research Report #7, Publications of the European Communities, van Wambeke, A.: Soils of the Tropics – Properties and Appraisal, Luxembourg, 2001. McGraw-Hill, New York, 1992. Shoji, S., Nanzyo, M., and Dahlgren, R.: Volcanic Ash Soils: Gen- van Wambeke, A., Eswaran, H., Herbillon, A. J., and Comera, J.: esis, Properties and Utlization, Developments in Soil Science 21, Oxisols, edited by: Wilding, L. P., Smeck, N. E. and Hall, G. F., Elsevier, 288, New York, USA. 1993. in: Pedogenesis and Soil Taxonomy: The Soil Orders, Elsevier, Schobbenhaus, C. and Bellizzia, A. (coord.): Geological Map of Amsterdam, 325–354, 1983. South America, 1:5,000,000, CGMW-CPRM-DNPM-UNESCO, Vitousek, P. M. and Sanford, R. L.: Nutrient cycling in moist tropi- Paris, 2001. cal forest, Annu. Rev. Ecol. Syst., 17, 137–167, 1986. Smeck, N. E.: Phosphorus dynamics in soils and landscapes, Geo- Went, F. W. and Stark, N.: Mycorrhiza, Bioscience, 18, 1035–1039, derma, 36, 185–199, 1985. 1968. Soares-Filho, B. S., Nepstad, D. C., Curran, L. M., Cerqueira, G. C., West, L. T., Beinroth, F. H., Summer, M. F., and Kang, B. T.: Ul- Garcia, R. A., Ramos, C. A., Voll, E., McDonald, E., Lefebvre, tisols: characteristics and impacts on society, Adv. Agron., 63, P., and Schlesinger, P.: Modelling conservation in the Amazon 179–236, 1998. basin, Nature, 440, 520–523, 2006. Wilding, L. P., Drees, L. R., and Nordt, L. C.: Spatial variability: Sombroek, W. G.: A reconnaissance of the soils of the Brazilian Enhancing the mean estimate of organic and inorganic carbon in Amazon region, Centre for Agricultural Publications and Docu- a sampling unit, in: Assessment Methods for Soil Carbon, edited mentation, Wageningen, 1966. by: Lal, R., Kimble, J. M. Follett, R. F., and Stewart, B. A., CRC Sombroek, W. G.: Soils of the Amazon Region,in: The Amazon: Press, 69–86, Florida, USA, 1983. Limnology and Landscape Ecology of a Mighty Tropical River Zinn, Y. L., Lal, R., Bigham, J. M., and Resck, D. V. S.: Edaphic and its Basin, edited by: Sioli, H. and Junk, W., Dordrecht, 521– controls on soil organic carbon retention in the Brazilian Cer- 535, 1984. rado: Texture and mineralogy, Soil Sci. Soc. Am. J., 71, 1204– Sombroek, W. G.: Amazon landforms and soils in relation to bio- 1214, 2007. logical diversity, Acta Amazonica, 30, 81–100, 2000.

Biogeosciences, 8, 1415–1440, 2011 www.biogeosciences.net/8/1415/2011/